US Army Corps of Engineers - Guidance For Evaluating Performance Based Chemical Data PDF

EM 200-1-10
30 Jun 05
ENVIRONMENTAL QUALITY
GUIDANCE FOR EVALUATING PERFORMANCE-

BASED CHEMICAL DATA
This manual is approved for public release; distribution is unlimited.
ENGINEER MANUAL
EM 200-1-10
30 Jun 05
AVAILABILITY
Electronic copies of this and other U.S. Army Corps of Engineers

(USACE) publications are available on the Internet at
http://www.usace.army.mil/inet/usace-docs/. This site is the only
repository for all official USACE engineer regulations, circulars,
manuals, and other documents originating from HQUSACE.
Publications are provided in portable document format (PDF).
vPrinted on Recycled Paper

DEPARTMENT OF THE ARMY EM 200-1-10
CECW-E U.S. Army Corps of Engineers
CEMP-RT Washington, DC 20314-1000
Manual
No. 200-1-10 30 June 2005
Environmental Quality
GUIDANCE FOR EVALUATING PERFORMANCE-BASED CHEMICAL DATA
1. Purpose. This Engineer Manual (EM) provides specific guidance, procedures, criteria,
and tools to improve the evaluation of chemical environmental data. In particular, the EM pre-
sents strategies to more effectively evaluate data in the context of its end use, which is referred
to as performance-based data evaluation. Performance-based data evaluation is recommended
to help ensure that only scientifically defensible data are used to support the decision making
process after project-specific data quality objectives have been establish. This EM is intended
for use by USACE personnel as a critical companion document to ER 1110-1-263.
2. Applicability. This EM applies to HQUSACE elements, major subordinate commands,

districts, laboratories, and field operating activities having responsibility for hazardous, toxic,
and radioactive waste (HTRW) projects. This includes, but is not limited to, execution of the
following programs: Defense Environmental Restoration Programs; Base Realignment and Clo-
sure; Superfund; Civil Works; Military Construction; installation environmental compliance; De-
fense Logistics Agency; Department of Energy; work for others; and any construction projects
involving hazardous, toxic, and radioactive waste.
3. Distribution Statement. Approved for public release and unlimited distribution.
4. References:
a. ER 1110-1-263
b. ER 1180-1-6
c. ER 5-1-11
d. ER 1110-1-12
EM 200-1-10
30 Jun 05
5. Discussion. This EM provides guidance for screening the usability of chemical data after
sample collection and analysis and for documenting the evaluation, in support of satisfying the
requirements of the USACE Environmental Quality Assurance (QA) Program for chemical data
as prescribed in ER 1110-1-263 and other general USACE quality management policy.
FOR THE COMMANDER:
2 Appendices
(See Table of Contents) Colonel, Corps of Engineers
Chief of Staff
DEPARTMENT OF THE ARMY EM 200-1-10
CECW-E U.S. Army Corps of Engineers
CEMP-RT Washington, DC 20314-1000
Manual
No. 200-1-10 30 June 2005
Environmental Quality
GUIDANCE FOR EVALUATING PERFORMANCE-BASED CHEMICAL DATA
Table of Contents
Subject Paragraph Page
Chapter 1. Introduction
Opening Remarks ........................................................................................................... 1-1 1-1
Scope and Limitations of Performance-Based Data Review.......................................... 1-2 1-1
Performance-Based Data Review Versus Data Validation ............................................ 1-2.1 1-3
Performance-Based Data Review Versus Usability Assessment ................................... 1-2.2 1-4
Overview of Performance-Based Data Review.............................................................. 1-3 1-5
Chapter 2. Data Review Reports

Introduction .................................................................................................................... 2-1 2-1
Cover Page ..................................................................................................................... 2-2 2-1
Cover Letter.................................................................................................................... 2-3 2-1
Executive Summary ....................................................................................................... 2-4 2-2
Technical Summary........................................................................................................ 2-5 2-2
Data Summary Tables .................................................................................................... 2-6 2-4
Project Specific Communications .................................................................................. 2-7 2-4
Project Specific Communications .................................................................................. 2-8 2-5
Chapter 3. Data Qualifiers

Introduction .................................................................................................................... 3-1 3-1
Definitions of Data Qualifiers ........................................................................................ 3-2 3-2
Chapter 4. Completeness
Introduction .................................................................................................................... 4-1 4-1
Minimum Reporting Requirements ................................................................................ 4-2 4-1
Cover Letter.................................................................................................................... 4-2.1 4-1
Case Narrative ................................................................................................................ 4-2.2 4-2
Technical Summary........................................................................................................ 4-2.3 4-2
Sample Management Records ........................................................................................ 4-2.4 4-4
Batch QC Summary Results........................................................................................... 4-2.5 4-4
Standard Preparation Logs ............................................................................................. 4-2.6 4-4
Sample Preparation Logs................................................................................................ 4-2.7 4-5
Instrument Run-Sequence Logs ..................................................................................... 4-2.8 4-5
Traceability..................................................................................................................... 4-2.9 4-6
Calibration Summary Results......................................................................................... 4-2.10 4-6
Chromatographic Methods for Organic Target Analytes ............................................... 4-2.11 4-7
Initial Calibration ........................................................................................................... 4-2.11.1 4-7
Internal Standard Summary Information........................................................................ 4-2.11.2 4-7
Surrogate Results............................................................................................................ 4-2.11.3 4-8
i
EM 200-1-10
30 Jun 05
Chromatographic Methods With 2-D Detectors............................................................. 4-2.11.4 4-8

Evaluation of Completeness ........................................................................................... 4-3 4-8
Missing Blanks ............................................................................................................... 4-3.1 4-9
Missing Laboratory Control Samples............................................................................. 4-3.2 4-10
Matrix Spike Data .......................................................................................................... 4-3.2.1 4-10
Surrogate Spike Data...................................................................................................... 4-3.2.2 4-11
CCV Data ....................................................................................................................... 4-3.2.3 4-11
Missing Detection Limits ............................................................................................... 4-3.3 4-11
Chapter 5. Holding Times and Preservation

Introduction .................................................................................................................... 5-1 5-1
Acceptance Criteria ........................................................................................................ 5-2 5-1
Evaluation....................................................................................................................... 5-3 5-1
Qualification................................................................................................................... 5-4 5-3
Low Stability Target Analytes........................................................................................ 5-4.1 5-3
High Stability Target Analytes....................................................................................... 5-4.2 5-3

Introduction .................................................................................................................... 6-1 6-1
Method Reporting Limits ............................................................................................... 6-2 6-1
Establishing Method Reporting Limits .......................................................................... 6-2.1 6-1
Qualification................................................................................................................... 6-2.2 6-2
Method Quantitation Limits ........................................................................................... 6-3 6-3
Establishing Method Quantitation Limits ...................................................................... 6-3.1 6-3
Qualification................................................................................................................... 6-3.2 6-5
Chapter 7. Initial Calibration

Introduction .................................................................................................................... 7-1 7-1
Acceptance Criteria ........................................................................................................ 7-2 7-1
Frequency ....................................................................................................................... 7-2.1 7-1
Number of Calibration Standards................................................................................... 7-2.2 7-1
Linear Calibration........................................................................................................... 7-2.3 7-2
Organic Methods ............................................................................................................ 7-2.3.1 7-2
Inorganic Methods.......................................................................................................... 7-2.3.2 7-3
Nonlinear Calibration ..................................................................................................... 7-2.4 7-3
Intercept of Calibration Curve........................................................................................ 7-2.5 7-3
Evaluation....................................................................................................................... 7-3 7-4
Goodness of Fit........................................................................................................... 7-3.1 7-4
Representativeness of Initial Calibration Curve............................................................. 7-3.2 7-4
Qualification................................................................................................................... 7-4 7-5
Frequency and Number of Standards ............................................................................. 7-4.1 7-5
Representativeness ......................................................................................................... 7-4.2 7-5
Goodness of Fit........................................................................................................... 7-4.3 7-5
Calibration Range........................................................................................................... 7-4.4 7-7
Chapter 8. Initial Calibration Verification (ICV)

Introduction .................................................................................................................... 8-1 8-1
Criteria............................................................................................................................ 8-2 8-1
Evaluation....................................................................................................................... 8-3 8-1
ii
EM 200-1-10
30 Jun 05
Qualification................................................................................................................... 8-4 8-2

Frequency ....................................................................................................................... 8-4.1 8-2
Percent Recovery............................................................................................................ 8-4.2 8-2
Inorganics ....................................................................................................................... 8-4.2.1 8-2
Organics.......................................................................................................................... 8-4.2.2 8-2
Qualification for Bias ..................................................................................................... 8-4.3 8-2
Chapter 9. Continuing Calibration Verification (CCV)

Introduction .................................................................................................................... 9-1 9-1
Criteria............................................................................................................................ 9-2 9-1
Traceability and Reporting Requirements...................................................................... 9-2.1 9-1
Frequency ....................................................................................................................... 9-2.3 9-1
Chromatographic Methods ............................................................................................. 9-2.3.1 9-2
Acceptance Criteria ........................................................................................................ 9-2.4 9-2
Organic Chromatographic Methods ............................................................................... 9-2.4.2 9-3
Evaluation....................................................................................................................... 9-3 9-3
Qualification................................................................................................................... 9-4 9-4
Frequency ....................................................................................................................... 9-4.2 9-5
Tolerance for Uncertainty............................................................................................... 9-4.3 9-5
Inorganic Methods, CCVs Not Processed with Samples ............................................... 9-4.3.1 9-6
Inorganic Methods, CCVs Processed with Samples ...................................................... 9-4.3.2 9-6
Organic Methods, CCVs Not Processed with Samples .................................................. 9-4.3.3 9-6
Organic Methods, CCVs Processed with Samples......................................................... 9-4.3.4 9-6
General Qualification Strategies..................................................................................... 9-4.4 9-6
Chapter 10. Blanks

Introduction .................................................................................................................... 10-1 10-1
Criteria............................................................................................................................ 10-2 10-1
Frequency ....................................................................................................................... 10-2.1 10-1
Acceptance Limits .......................................................................................................... 10-2.2 10-1
Evaluation....................................................................................................................... 10-3 10-2
Contractual Considerations ............................................................................................ 10-4 10-2
Qualification for Blank Contamination .......................................................................... 10-5 10-3
Chapter 11. Laboratory Control Samples (LCSs)

Introduction .................................................................................................................... 11-1 11-1
Criteria............................................................................................................................ 11-2 11-1
Frequency ....................................................................................................................... 11-2.1 11-1
Evaluation....................................................................................................................... 11-3 11-2
Qualification................................................................................................................... 11-5 11-4
Qualification Strategies Using Estimates of the Uncertainty ......................................... 11-6 11-6
iii
EM 200-1-10
30 Jun 05
Chapter 12. Matrix Spikes, Matrix Spike Duplicates, and Matrix Duplicates
Introduction .................................................................................................................... 12-1 12-1
Interpretation of Matrix Spike and Duplicate Results .................................................... 12-2 12-1
Estimating Statistical Matrix Spike Recovery Ranges ................................................... 12-3 12-3
Criteria............................................................................................................................ 12-4 12-4
Frequency ....................................................................................................................... 12-4.2 12-5
Project Specific Communications .................................................................................. 12-4.3.1 12-5
Establishing Acceptance Limits for Matrix-Dependent Duplicates ............................... 12-4.3.2 12-7
Evaluation....................................................................................................................... 12-5 12-7
Qualification................................................................................................................... 12-7 12-8
Matrix Spikes and Matrix Spike Duplicates................................................................... 12-7.1 12-8
Matrix-Dependent Duplicates ........................................................................................ 12-7.2 12-10
Chapter 13. Surrogates

Introduction .................................................................................................................... 13-1 13-1
Criteria............................................................................................................................ 13-2 13-1
Evaluation....................................................................................................................... 13-3 13-2
Qualification................................................................................................................... 13-5 13-4

Introduction .................................................................................................................... 14-1 14-1
Criteria............................................................................................................................ 14-2 14-1
Frequency ....................................................................................................................... 14.2.1 14-1
Acceptance Limits .......................................................................................................... 14.2.2 14-1
Evaluation....................................................................................................................... 14-3 14-2
Qualification................................................................................................................... 14-4 14-2
MS Recovery Acceptable and PDS Recovery Unacceptable ......................................... 14-4.1 14-2
MS Recovery Unacceptable and PDS Recovery Acceptable ......................................... 14-4.2 14-3
MS Recovery Unacceptable and PDS Recovery Unacceptable ..................................... 14-4.3 14-3

Introduction .................................................................................................................... 15-1 15-1
Criteria............................................................................................................................ 15-2 15-1
Frequency ....................................................................................................................... 15-2.1 15-1
Duplicate Precision......................................................................................................... 15-2.2 15-1
Evaluation....................................................................................................................... 15-3 15-2
Qualification................................................................................................................... 15-4 15-2
Chapter 16. Internal Standards for Organic Chromatographic Methods

Introduction .................................................................................................................... 16-1 16-1
Criteria............................................................................................................................ 16-2 16-1
Frequency ....................................................................................................................... 16-2.1 16-1
Acceptance Criteria ........................................................................................................ 16-2.2 16-1
Evaluation....................................................................................................................... 16-3 16-2
Qualification................................................................................................................... 16-4 16-2
iv
EM 200-1-10
30 Jun 05
Appendix A
Glossary................................................................................................................................... A-1
Appendix B
Holding Times and Preservation ............................................................................................B-1
v
EM 200-1-10
30 Jun 05
CHAPTER 1
Introduction
1-1. Opening Remarks.
a. The production of data of known and acceptable quality is a primary goal of every
environmental restoration and compliance sampling effort. In general, some degree of data re-
view should be performed for all data collection activities to help ensure that only scientifically
defensible data are used to support project decisions. However, the extent of the review will be
dependent upon the projects data quality objectives (DQOs) and will be limited by the physi-
cal contents of the data package.1 For example, the reporting and evaluation requirements for
definitive data and screening data will differ significantly.
b. This document provides guidance to the U.S. Army Corps of Engineers (USACE) and
USACE contractors (e.g., to architect-engineering contractors and third-party data reviewers) for
evaluating instrumental chemical data using a performance-based approach. A performance-
based method is defined as an analytical procedure for which data quality indicators are
documented and evaluated with respect to acceptance criteria that are established from project
data quality objectives. In particular, the PARCCS parameters (i.e., precision, accuracy, com-
pleteness, representativeness, comparability, and sensitivity) are documented for the target
analytes of concern at the levels of concern (i.e., at or below project action levels) in the envi-
ronmental media of interest and are evaluated with respect to acceptance limits or measurement
quality objectives (MQOs) that are designed to ensure that total measurement uncertainty is
within the limits prescribed by project DQOs. This document assumes that DQOs and MQOs
have been established and presents guidance for evaluating chemical data quality, as measured
by PARCCS, as a first step process for data usability assessment. (Refer to Chapter 1.2.2 for
additional discussion regarding data usability assessment.) 2 To more effectively assess data us-
ability, it is recommended that existing data evaluation protocols and checklists be revised using
at least some of the strategies presented in this document.
1-2. Scope and Limitations of Performance-Based Data Review.
a. In general, data packages must contain enough information to evaluate PARCCS.3

Data packages must essentially contain summary results for instrument calibrations (initial and
continuing), environmental samples, and associated batch quality control (QC) samples (e.g.,
method blanks and laboratory control samples), as well as select raw data. (Specific reporting
requirements are addressed in Chapter 4 of this document.) Data packages that satisfy these re-
porting requirements will be referred to as performance-based (PB) data packages and the im-
1
The definitions of a number of critical terms appear in bold print the first time they are used. These terms are
defined in the body of the document or, more commonly, in the glossary at the end of the document. Italics are
used to denote emphasis or special meaning.
2
Since it is assumed that MQOs have been established and are consistent with DQOs, for simplicity, from hence
forth, the distinction between the two terms will not be maintained; DQOs will also refer to the MQOs.
3
Although this document constitutes guidance, terms such as must, shall, and will are used in the document
when an item is viewed to be especially critical or when an activity is viewed to be typically appropriate.
1-1
EM 200-1-10
30 Jun 05
plementation of the data evaluation activities described this document will be referred to as per-
formance-based (PB) data review.
b. The data review protocols presented in this document should not be viewed as
prescriptive algorithms but as strategies intended for the purposes of guidance. For example,
quality control (QC) acceptance limits are specified in this document but these limits should be
viewed as baseline limits that should be adjusted (i.e., increased or decreased) based upon the
objectives of the project. Even if it were possible to specify a set of QC acceptance limits that
would be applicable to all projects, the potential occurrence of multiple QC problems alone
suggests that a prescriptive approach for data evaluation would be unfeasible (e.g., it would not
be practical to propose an evaluation strategy for every combination of QC problems that could
be encountered). Because of the complexities of environmental investigations and uniqueness of
environmental samples, analytical data must ultimately be evaluated using professional
judgement in the context of project-specific data objectives.
c. In order to successfully implement a PB review, the DQOs in project planning docu-

ments such as Work Plans, Sampling and Analysis (SAPs), and Quality Assurance Project Plans
(QAPPs) must be well defined. The generation of DQOs is beyond the scope of this document.
However, it should be noted that generic statements such as definitive or Level IV data will
be collected will not suffice. A specific set of QC acceptance limits must be presented for the
analytes of concern for the concentrations of interest for the environmental populations being
sampled. In theory, project documents such as QAPPs contain comprehensive and appropriate
QC specifications, but, in practice, this is not necessarily true (e.g., when the data reviewer is not
adequately involved in the project planning process).
d. In order to perform a PB data review, project-specific data quality objectives must be

scientifically defensible. The scientific defensibility of the data should take precedence over
contract compliance issues or the QAPP when QAPP contains inappropriate specifications. For
example, if the QAPP requires data to be evaluated solely upon the basis of method-specified
QC criteria (e.g., such as those specified in SW-846 methods) or laboratory performance criteria,
then the validity of assessing the data on this basis should be carefully evaluated before pro-
ceeding with the data review. In particular, sensitivity requirements should not be established
solely on the basis of method-specified quantitation limits such as the Contract Laboratory Pro-
gram (CLP) Contract Required Quantitation Limits (CRQLs). Method-specified quantitation
limits may be inappropriately high for project-specific action levels (e.g., risk-based cleanup lev-
els). Laboratory detection, quantitation, and reporting limits must be evaluated with respect to
the project-specific action levels to demonstrate that the proposed analytical methods possess
adequate sensitivity. Similarly, it would typically be inappropriate for the QAPP to establish
method data quality objectives for precision and bias solely on the basis of a laboratorys statis-
tical control limits (e.g., for matrix spikes and laboratory control samples). A laboratorys sta-
tistical limits may be indicative of the laboratorys routine performance but may be too wide to
yield quantitatively reliable results.
e. Performance-based data review must not be performed as a last-minute activity

that is initiated only after the completion of all sample collection and analysis. To the extent
that is possible or practical, prior to performing a PB data review, the reviewer should possess a
1-2
EM 200-1-10
30 Jun 05
complete understanding of the intended use of the data and the relationship of the QC results to
the usability of the data. The reviewer must receive input from the end-data users regarding the
objectives and expected results of the analyses (e.g., via the review of the Project Quality Assur-
ance Plan and Sampling and Analysis Plan). For optimal results, the reviewer should be in-
volved in the DQO process in the early planning stages of the project (e.g., should be involved in
scoping meetings where project DQOs, scheduling, sampling techniques, analytical methodolo-
gies, and data evaluation criteria are established.) When the data reviewer is not adequately in-
volved in the DQO process, a PB data review may result in the rejection of a significant portion
of the analytical data. Performance-based data evaluation strategies need to be specified during
project planning
1-2.1. Performance-Based Data Review Versus Data Validation.
a. This guidance is generally applicable to any instrumental performance-based method,

regardless of the determinative or preparatory techniques used to process the environmental
samples. The data review protocols will result in a relatively thorough evaluation of data quality
and will be applicable to a variety of environmental projects. However, the data review strate-
gies presented in this document may be insufficient for all data uses. Project-specific DQOs may
require more comprehensive data evaluation activities than those performed during PB data re-
view.
b. Performance-based data review does not constitute data validation. Data validation
is a more in-depth evaluation of laboratory data quality and is beyond the scope of this docu-
ment. As the term is used in this document, data validation refers to any independent systematic
review of comprehensive data packages with respect to a predefined set of technical perform-
ance criteria for PARCCS. A comprehensive data package is defined as a data package that
contains sufficient information to completely reconstruct the laboratory analyses that were per-
formed and documents salient field sample collection and handling activities (e.g., contains the
Chain-of-Custody and may contain field logs). Hence, comprehensive data packages contain
summary data for environmental, batch QC, and instrument QC sample analyses as well as all
the raw laboratory data (e.g., standard preparation logs and printouts of chromatograms). CLP
data packages are examples of comprehensive data packages with distinct reporting require-
ments.
c. Data validation involves the evaluation of batch QC and calibration results, in addition
to other instrument QC results using the raw data. Since all the raw laboratory data are not in-
cluded in performance-based data packages (unlike for data validation), reported QC summary
results (e.g., laboratory control sample and surrogate recoveries) are not verified to the level of
the raw data (e.g., using chromatograms and other instrumental printouts). Furthermore, with the
exception of calibration data, PB data packages do not contain instrument QC results (e.g., pesti-
cide percent breakdown and tune checks). Hence, during PB data review, instrument perform-
ance (other than calibration) is assumed to be in control or out-of-control in a manner that is con-
sistent with batch QC performance. This assumption is usually reasonable but is not always
valid.
1-3
EM 200-1-10
30 Jun 05
d. A more thorough data evaluation should be considered when significant QC problems

are observed during PB data review or when data is being collected to support critical deci-
sions. Since laboratories normally maintain files of all supporting data and documentation for
the analyses performed (for the period of time that is normally specified in the contract for ana-
lytical services), the laboratory can be requested to provide copies of the raw data to perform a
more comprehensive review when the need arises. However, it is recommended that require-
ments for archiving comprehensive data packages be explicitly addressed when contracting for
analytical services.
e. During project planning, the objectives of the analyses, nature of the contamination,
limitations of the analytical methodology, and historic information about the site should be
evaluated to determine whether a more comprehensive review needs to be performed. In par-
ticular, if the analytical technique involves the use of a 2-D detector rather than a 3-D detector,
then it is especially critical to take stability problems (e.g., photochemical and thermal degrada-
tion) and interferences into account when determining whether or not a more comprehensive
evaluation is required. For example, a review of batch QC results alone would probably be in-
adequate to identify data quality problems when a high performance liquid chromatograph
(HPLC) with a fluorescence detector is being used to measure low levels of polynuclear aromatic
hydrocarbons (PAHs) such as benzo(a)pyrene at a site with high background fuel contamination.
The evaluation of 4,4'-DDT and Endrin breakdown checks (e.g., as discussed in Method 8081A)
may be required to determine whether or not detections of target analytes such as Endrin ketone
and Endrin aldehyde are actually false positives arising from poor method implementation (e.g.,
the degradation of Endrin during instrumental analysis).
Note: The evaluation strategies presented in this document may be less adequate for 2-D
detector methods than 3-D detector methods. However, this does not imply that the strate-
gies are not appropriate or useful for 2-D detector methods. The level of confidence for
data will be a function of the nature of the analytical technique, regardless of the thorough-
ness of any data evaluation activity. It is merely being noted that, since 2-D methods in-
herently lack the specificity of 3-D methods, 2-D methods are more prone to data quality
problems (e.g., false positives) that, under select circumstances, may only be identified via
the evaluation of a full raw data package.
1-2.2. Performance-Based Data Review Versus Usability Assessment.
a. It is emphasized that the PB data review activities discussed in this document

constitute only a first-step process for the assessment of data usability. A full assessment of data
usability is a more complex and comprehensive activity than PB data review or validation; the
former encompasses the latter and is potentially more subjective. The data user must ultimately
assess the overall usability of data on the basis of total measurement uncertainty and the
objectives of the investigation. .
b. Total measurement uncertainty consists of the sum of the laboratory analytical uncer-
tainty and field sampling uncertainty. Unfortunately, field sampling uncertainty is often greater
than laboratory analytical uncertainty and is not fully taken into account during data review or
validation (i.e., data review and validation identify laboratory analytical uncertainty but do not
1-4
EM 200-1-10
30 Jun 05
fully address field sampling uncertainty). For example, data review and validation may identify
incorrect preservation techniques but would not adequately characterize the representativeness of
a sample collected from an environmental population with high spatial or temporal variability.
c. Little or no usability assessment is typically performed during data review or valida-

tion. Usability assessment is usually performed after data review or validation is completed. For
example, when data validation is performed using the National Functional Guidelines, sensitivity
is evaluated with respect to fixed CRQLs rather than project-specific action levels. However,
meeting CRQLs does not ensure that the data will be usable (a problem which, unfortunately,
many usability assessments also fail to identify). This document constitutes a more streamlined
approach for data review. Data quality is evaluated during data review in the context of the end
use of the data.
1-3. Overview of Performance-Based Data Review.
a. This section of the document presents a brief overview of the PB data review process.
The reviewer initially receives input from the end-data users regarding the objectives and ex-
pected results of the analytical efforts (e.g., in the form of formal DQOs described in the Work
Plan and QAPP). Prior to performing a PB data review, the reviewer performs a cursory evalua-
tion of the data package to ensure that it contains all the required documentation. This is critical
since the evaluation of any data package will be limited by its physical content. If the data pack-
age is essentially complete, the reviewer performs a more complete evaluation to determine if the
data potentially meet the needs of the end user. The reviewer verifies that sample collection and
handling activities were properly implemented in the field and subsequently evaluates the ana-
lytical quality of the laboratory data. A PB review includes the evaluation of the following QC
elements:
(1) Completeness.
(2) Holding Time and Preservation.
(3) Initial Calibration.
(4) Initial Calibration Verification.
(5) Continuing Calibration Verification.
(6) Sensitivity (e.g., detection and quantitation limits).
(7) Blanks (e.g., field and method blanks).
(8) Laboratory Control Samples.
(9) Post-Digestion Spikes (for trace metal methods).
(10) Matrix Spikes.
1-5
EM 200-1-10
30 Jun 05
(11) Matrix Spike Duplicates and Matrix Duplicates.
(12) Surrogates (for organic chromatographic methods).
b. Detailed definitions of these QC elements may be found in the USACE Shell for
Analytical Chemistry and the glossary of this document.
c. Quality control samples are designed to evaluate the PARCCS parameters and identify
quality problems in three specific areas: (i) Laboratory analytical performance, (ii) matrix ef-
fects, and (iii) field performance. For example, accuracy is assessed from calibration, laboratory
control sample (LCS), matrix spike (MS), post-digestion spike (PDS), and surrogate data. Preci-
sion is evaluated from duplicate laboratory control and matrix spike samples. Sensitivity is
evaluated using detection limits and quantitation limits. Representativeness is evaluated via the
review of holding time and blank data. A laboratorys analytical performance is evaluated using
calibration results (i.e., initial calibrations, initial calibration verifications, and continuing cali-
bration verifications) and batch QC samples such as method blanks and laboratory control sam-
ples. Matrix effects are evaluated using matrix spike, surrogate spike, and post digestion spike
recoveries. Field duplicates, rinsate blanks, and trip blanks are examples of QC samples that
are used to assess QC problems associated with sample collection activities..
d. After (or during) the technical evaluation, the reviewer generates a data review re-
port that summarizes the overall quality of the data and lists individual QC problems and any
observations that may be relevant to the datas potential usability. Data review reports are dis-
cussed in Chapter 2.
1-6
EM 200-1-10
30 Jun 05
CHAPTER 2
Data Review Reports
2-1. Introduction.
A data review report documents the PB data evaluation. At least one data review report must be
generated for each data sample delivery group. The data review report may address the data
packages for several analytical methods. The format of the document is not as important as its
content. However, a uniform format is recommended to facilitate data evaluation activities. The
elements presented below must be included in a data review report.
2-2. Cover Page.
The cover page specifies the following information:
a. Unique report ID number.
b. Name and address of data reviewer.
c. Contract number.
d. Client name and address.
e. Project name and site location.
f. Statement of data authenticity and official signature of release.
2-3. Cover Letter.
a. Project name (or brief description of the project).
b. Site name (location from which the samples were collected).
c. Parties responsible for evaluating the data and the date the evaluation was performed
(including a point of contact for questions with phone and facsimile numbers).
d. Technical criteria used to evaluate the data (e.g., cited as a reference).
e. Laboratory that performed the analyses (name, location, and point of contact).
f. Description of the samples that were evaluated, including the following:
(1) Number of samples.
(2) Matrix.
2-1
EM 200-1-10
30 Jun 05
(3) Environmental samples associated with the field QC samples.
(4) Field and laboratory ID numbers.
(5) Date samples were collected in the field.
(6) Preparatory and determinative methods of analysis (including method numbers).
(7) Target analytes or parameters.
(8) Date the laboratory analyses were performed.
(9) Date the data package from the laboratory was received.
2-4. Executive Summary.
a. The objective of the Executive Summary is to concisely describe the overall quality of
the data package in a manner that is comprehensible to an individual lacking an extensive back-
ground in analytical chemistry. Major areas of concerns and any information which would aid
the reader to better understand the quality or usability of the data must be discussed in general
terms. For example, Executive Summary may state that a complete review of the data could not
be performed because of missing information or it may state that no significant QC problems
were observed.
b. When major QC problems are observed after the data review process, when possible,
the Executive Summary must indicate whether these problems primarily resulted from unaccept-
able laboratory performance, matrix interference, or problems associated with the sample col-
lection activities. If the QC problems resulted because the laboratory or field personnel failed to
follow the requirements in the Work Plan or QAPP, this information should be highlighted. The
Executive Summary should also recommend corrective actions to improve the quality of the
data. The format and content of the Executive Summary are otherwise left to the discretion of
the author.
2-5. Technical Summary.
a. The Technical Summary must discuss the quality of the data package in terms of spe-
cific QC elements and must be divided into subsectionsone for each QC element in which
problems were identified (e.g., Holding Times, Laboratory Control Samples, Matrix
Spikes, and Continuing Calibration Verifications). The Technical Summary must discuss the
effects of QC problems in the context of sensitivity (e.g., false negatives due to high detection
limits), precision (e.g., high variability), accuracy (e.g., high or low bias), representativeness
(e.g., blank contamination), completeness (e.g., missing information), and comparability (e.g.,
failure to use specified methodology).
2-2
EM 200-1-10
30 Jun 05
b. The Technical Summary explain why each result was qualified (e.g., matrix interfer-
ence and blank contamination). This is especially critical when project DQOs are not well de-
fined for a particular parameter or when qualification is based upon a high degree of professional
judgment (e.g., due to the complexity of the projects objectives or the environmental population
being sampled). Any changes made to the laboratorys reported data (e.g., due to misiden-
tification, transcription errors, or calculation errors) must be identified and the samples affected
by each QC problem should be listed in a tabular format.
c. The Technical Summary must identify QC problems as having a major or minor

impact on data quality or usability. The Technical Summary should also distinguish systematic
errors from random errors. Systematic errors resulting from blunders (e.g., transcription errors)
should also be distinguished form systematic effects that bias the results (e.g., from poor
extraction efficiency). When possible, the Technical Summary should identify the direction of
bias (high or low). In addition, problems giving rise to qualitative uncertainties must be
distinguished from those giving rise to quantitative uncertainties. Qualitative uncertainty refers
to uncertainty associated with the identification of an analyte in an environmental sample.
Quantitative uncertainty refers to error associated with the determination of the amount of an
identified analyte. (Refer to the definitions of the N and J qualifiers in Chapter 3.)
d. When possible, field sampling uncertainty must be distinguished from laboratory

analytical uncertainty. Problems arising from missing data and QC failures resulting from sub-
standard laboratory performance (e.g., out-of-control LCS recoveries) and substandard sample
collection procedures (e.g., the lack of sample preservation) must be highlighted. When major
QC problems are observed, corrective actions should be recommended. However, it should be
noted that the major of objective of the evaluation is to determine the potential usability of the
data and not contractual compliance (e.g., contractually noncompliant data may or may not be
usable.)
e. The reviewer should avoid statements pertaining to the ultimate usability of the data.
As defined in this document, PB data review results in usability screening rather than a full
usability assessment. In particular, unless there is a high degree of confidence that a set of re-
sults must be rejected (e.g., the results are being qualified with the R flag), adjectives such as
unusable and unacceptable should be avoided (e.g., the results should be described as ten-
tatively unusable.) The use of these terms in data review reports may be interpreted as contra-
dictory by regulators in situations where the end users determine that the data are useful for pro-
ject purposes in spite of the QC problems. Similarly, terms such as usable, acceptable, and
valid, should only be used when the report explicitly defines these terms to mean that the data
are potentially usable or that the data review specifications have been satisfied. Examples of
preferred terminology are presented below.
(1) The results for the aqueous HVOCs (laboratory batch 50603, samples SL5-3031-1
to SL-3031-6) may possess a negative bias because the samples were analyzed one day beyond
the holding time limit; detections are qualified with the J- flag (i.e., as estimated with suspected
low bias) and nondetections are qualified with the UN flag to indicate the possibility of false
negatives ....
2-3
EM 200-1-10
30 Jun 05
(2) The Chain-Of-Custody was not signed in the field. This may adversely impact the
legal defensibility of the data. However, no results were qualified based upon this observation
...
(3) The low LCS recovery (11%) for the semivolatile pentachlorophenol for (labora-
tory) Batch 49382 is indicative of a large negative bias for the associated samples (GW-2-21-
972, GW-2-21-972-FD, GW-2-21-972-MS, and GW-2-21-973 to GW-2-21-980). Detected con-
centrations of the analyte are considered to be minimum values and nondetections are considered
to be unreliable at the stated reporting limits. Reported concentrations (nondetections and de-
tections) of the analyte below the project-specified action level are qualified with the X flag as
tentatively unusable because they do not demonstrate that the analyte is present below the action
level. Detections above the action level are qualified with the J- flag ...
2-6. Data Summary Tables.
Present qualified results for the environmental samples in a tabular format and list the definitions
of all data qualifiers. Use footnotes to briefly explain why the data were qualified. The sum-
mary tables should also list the detection and quantitation limits and any project-specific action
levels or requirements for sensitivity. (It is recommended that this be done in a format that will
enable the data to be readily exported to the project report.) The header information for each ta-
ble typically includes the following information:
a. Project name and location.
b. Laboratory name and location (City and State).
c. Field and laboratory ID numbers.
d. Matrix type.
e. Preparatory and determinative method.
f. Date of sampling, analysis, and preparation.
g. Amount of sample processed and analyzed (including extract volume).
h. Dilution factors.
i. Percent moisture (for solid samples).
j. Concentration units.
2-7. Project Specific Communications.
This section contains pertinent communications (e.g., phone logs, E-mail, and letters) between
the data reviewer and any agencies or parties that possess an interest in the quality of the data
2-4
EM 200-1-10
30 Jun 05
(e.g., the analytical laboratory, the contractor that collected the samples, the USACE district
field office, and regulators). Examples are listed below:
a. Requests to the analytical laboratory for the submittal of additional information.
b. Communications with the client concerning major data quality deficiencies.
c. Communications with the samplers to address QC problems associated with sample

collection.
d. Inquiries from regulatory authorities.
e. Requests from the client for quick turnaround time.
f. Amendments of the data quality objectives from the client.
2-8. Project Specific Communications.
a. Include checklists that were used to review the data packages in an appendix of the
data review report. Checklists demonstrate that the data packages were assessed for overall
completeness prior to the technical evaluation and appropriate QC elements were assessed
during the technical evaluation (e.g., holding times, initial calibration, and laboratory control
samples).
b. Include worksheets that were used to verify the laboratorys reported results (e.g., any
recalculations that were performed). When errors are observed, photocopies of the laboratorys
original data and any relevant field documents should be used to illustrate the corrections per-
formed. For example, if incorrect concentration units were reported for all the samples, it would
only be necessary to illustrate the correction for one sample. Note that, depending upon the se-
verity of the errors and the contractual requirements for the analyses, the laboratory may be re-
quired to correct the results and resubmit the data packages.
2-5
EM 200-1-10
30 Jun 05
CHAPTER 3
Data Qualifiers
3-1. Introduction.
a. Data qualification is an integral component of data review and validation. During

PB data review, data qualifiers or flags are applied to alert the end user to quality problems that
may impact the usability of the data (e.g., QC acceptance limits that were not met). However, it
is emphasized that data qualification essentially results in a qualitative evaluation of the data
(e.g., measurement uncertainty is not evaluated in a quantitative manner). This is one of the
major reasons why data review or validation is only a first-step process for the assessment of
data usability (as stated in Chapter 1.2.2).
b. During PB data evaluation, results are either accepted or reported with data qualifiers
or flags. Data that meet all QC acceptance limits are potentially usable and are not qualified.
Data that fail one or more QC criteria are qualified as estimated (with the J flag), tentatively re-
jected (with the X flag), or rejected (with the R flag). The distinction between estimated, tenta-
tively rejected, and rejected data resides in the degree of the QC failure and is highly dependent
upon the reviewers understanding of the objectives of the project.
c. In general, data that are believed to be completely unusable with a high degree of
confidence (e.g., because of the gross failure of QC criteria) are qualified as rejected and would
not normally be used to support decisions for an environmental project. Data associated with a
marginal failure of QC criteria that are believed to be tentatively usable or more usable than
not are qualified as estimated. Data that are mostly unusable or that fall into the gray area
between estimated and rejected are qualified as tentatively rejected.
Note: Typically, when data validation is performed, data are primarily qualified as ei-
ther estimated (e.g., with the J flag) or rejected (e.g., with the R flag). However, since
data are usually rejected only for the most severe or blatant QC problems, the R flag is
rarely applied. When QC problems are observed, the data are frequently qualified as
estimated and are subsequently used to support project decisions. Unfortunately, J-
qualified data are often used to support project decisions without evaluating the impact
of the QC problems on the usability of the data, resulting in an over estimation of data
quality. To minimize the potential indiscriminate use of J-qualified data, an additional
data qualifier, the X flag, has been defined. During a PB review, depending upon the
severity of the QC problem, data are primarily qualified as either estimated and tenta-
tively accepted (J flag), estimated and tentatively rejected (X flag) or rejected (R
flag).
d. As stated earlier, full data usability assessment is a more complex and comprehensive
activity than data review or validation and is usually performed by the end user (rather than by
the data reviewer) because the data user typically possesses a greater understanding of the pro-
jects DQOs (e.g., because of a more extensive knowledge of the projects history). Therefore,
the end user must ultimately determine the acceptability of the data. However, this does not im-
3-1
EM 200-1-10
30 Jun 05
ply that the end user may apply qualified data in an indiscriminate fashion. This is particularly
true of data that have been qualified as tentatively rejected. Tentatively rejected data must not be
used to support project decisions unless the data user presents (i.e., documents) some technical
rationale for doing so. In other words, tentatively rejected data must ultimately be rejected (e.g.,
using the R flag) in the absence of a scientifically defensible rationale to do otherwise. This re-
quirement should be explicitly specified in the QAPP. Furthermore, when data qualified as ten-
tatively rejected are used to support decisions for a project, the data reviewer should be con-
sulted for a consensus unless it is clear that the reviewer did not possess a complete
understanding of the objectives of the investigation (e.g., new DQOs were established after the
data review was performed). It should be noted Chemical Data Quality Assessment Reports
(as defined by EM 200-1-6, 10 October 1997) by USACE project chemists represent one possi-
ble mechanism to document a more comprehensive usability evaluation (e.g., X-flagged data
may be converted to J-flagged or R-flagged data in Chemical Data Quality Assessment Reports).
e. Ideally, estimated (i.e., J-qualified) data, though presumed to be usable by the data re-
viewer, should be accepted by the end user only after the reasons for the data qualifications and
their impact on the achievement of project DQOs have been examined. For example, when the
direction of bias and the magnitude of the analytical uncertainty are well defined, a more thor-
ough examination of the data may entail an evaluation similar to that presented in Paragraph 11-
6.
3-2. Definitions of Data Qualifiers.
a. All data qualifiers or flags must be clearly defined. Project-specific requirements ulti-
mately determine the types of qualifiers that are required (e.g., the EPA Functional Guidelines
for validation require a distinct set of flags). However, in the absence of more appropriate con-
ventions for data qualification, the flags defined below must be used. The definitions of the data
qualifiers are summarized in Table 3-1.
(1) R flag.
(a) The datum is rejected. The qualifier typically indicates that a datum is completely
unusable because it is of unknown quality (e.g., missing QC information) or because of gross QC
deficiencies (e.g., extremely poor recoveries for the LCS).
(b) NFor gas chromatography/mass spectrometry (GC/MS) analyses, the R flag must be
used to reject TIC tentatively identified compound results that are believed to be laboratory arti-
facts (i.e., common laboratory contaminants). Examples include reagent contaminants, solvent
preservatives, siloxanes, and aldol condensation reaction products of acetone (e.g., 4-hydroxy-4-
methyl-2-pentanone, 4-methyl-2-penten-2-one, and 5,5-dimethyl-2(5H)-furanone).
(2) J flag. The target analyte is positively identified, but the reported numerical result
(e.g., analyte concentration) is an estimated value and the direction of bias is unknown. The flag
indicates that a significant quantitative (as opposed to a qualitative) uncertainty exists. The J
flag must always be used to report the following.
3-2
EM 200-1-10
30 Jun 05
(a) Detections below the method quantitation limit (MQL).
(b) Tentatively Identified Compounds (TICs).1
(3) J-Flag. The target analyte is present but the reported numerical result is an estimate
that is believed to be biased low (e.g., the actual concentration in the environmental sample is
believed to be greater than the reported concentration).
(4) J+ Flag. The analyte is present but the reported numerical result is an estimate that
is believed to be biased high (e.g., the actual concentration in the environmental sample is
believed to be less than the reported concentration).
(5) N Flag. The target analyte is reported as a tentative detection (e.g., because the
identity of the analyte is in doubt). The N flag indicates a significant qualitative rather than
quantitative uncertainty exits (i.e., the reported detection of the analyte may be a false
positive). When used in combination with the U flag (i.e., the UN flag), the qualifier indicates
that the absence of a target analyte at some stated reporting limit is in doubt (i.e., a false negative
is possible at the stated reporting limit). Applications of the N flag include the following:
(a) Uncertain Aroclor identifications (e.g., weathered PCBs).
(b) Tentatively Identified Compounds.
(6) U Flag. The analyte was not detected relative to the method reporting limit
(MRL); that is, the result is less than the method reporting limit (MRL).
(7) NJ Flag. The presence or identity of the analyte is in doubt and the reported
concentration is estimated. The estimation is both qualitative and quantitative in nature.
(8) UN Flag. The result is reported as a tentative nondetection (as opposed to tentative
detection); there is uncertainty with whether or not the nondetection is valid at the stated method
reporting limit (e.g., because of QC problems).
Note: The UN flag is similar (but not identical) to the CLP UJ flag, which is defined
as follows: The analyte was not detected above the reported sample quantitation limit.
However, the reported quantitation limit is approximate and may or may not represent
the actual limit of quantitation necessary to accurately and precisely measure the ana-
lyte in the sample. Both flags indicate that an uncertainty is associated with a nonde-
tection. However, the UJ and UN flags differ in that the former is defined in terms of
the CLP CRQLs, while the later is defined in terms of project-specific reporting limits.
The UN and UJ qualifiers are essentially equivalent when the reporting limits equal the
CRQLs. However, in general, setting the reporting limits equal to the quantitation lim-
its is not recommended (e.g., unless the action levels for the project are high).
1
TICs are typically qualified by the laboratory.
3-3
EM 200-1-10
30 Jun 05
(9) X Flag. The datum is tentatively rejected because project-specific data quality objec-
tives (e.g., for sensitivity, accuracy, or precision) were not met or were not demonstrated. When
objectives for sensitivity are not met, the X flag typically indicates that a result (a detection or
nondetection) is potentially unusable with respect to an action level (e.g., the result does not
demonstrate that a target analyte is actually present in an environmental sample at a concentra-
tion above or below a risk-based decision limit).
Note: When evaluating objectives for sensitivity, the R flag may be more appropriate
than the X flag when action levels are fixed and statistical analyses are not being per-
formed. The X flag may be appropriate when action levels are subject to change, a set
of data is being evaluated with respect to different action levels, or when statistical
analyses are being performed.
b. It may be desirable to use the X flag in combination with other flags as illustrated be-
low:
(1) X- Flag. The detection is (quantitatively) grossly estimated with low bias and is
tentatively rejected.
(2) X+ Flag. The detection is (quantitatively) grossly estimated with high bias and
(3) XN Flag. The detection is quantitatively and qualitatively highly estimated and is
(4) XU Flag. The nondetection is tentatively rejected.
c. When a datum is qualified, the reviewer must explain why the particular qualifier was
applied. It is recommended that numerical subscripts be placed on flags to indicate why the flag
was used (e.g., the flags J1 and J2 may indicate that a result is estimated because of poor target
analyte recovery for the associated laboratory control sample and matrix spike, respectively).
However, the use of subscripts may not be practical for projects that involve a large number of
samples when the samples possess multiple QC problems. Similarly, it may be desirable to sup-
press the numerical subscripts when the reasons why the flags were applied are not important to
a particular data user. However, the rationale for each qualification must be explained in the
data review report.
d. When possible, the J flag must identify any suspected bias (high or low) in the data. If
bias is known for an estimated result, use either the J- or J+ flag to qualify a result. However, if
a datum is estimated because of multiple QC problems and the direction of bias is not well de-
fined, it may be appropriate to qualify the datum as tentatively rejected rather than as estimated.
3-4
EM 200-1-10
30 Jun 05
Table 3-1
Summary of Major Data Qualifiers
Qualifier Definition
J Estimated (quantitatively) and tentatively usable
J- Estimated (quantitatively) with low bias
J+ Estimated (quantitatively) with high bias
U Below reporting limit
N Qualitatively estimated (tentative detection)
X Tentatively rejected
R Rejected
UN Tentative nondetection
NJ Quantitatively and qualitatively estimated
3-5
EM 200-1-10
30 Jun 05
CHAPTER 4
Completeness 1
4-1. Introduction.
The primary objective of this review is to ensure that the data package contains adequate docu-
mentation to perform the PB data review. The deliverables that constitute a PB data package are
discussed below. Before performing a technical review of the data package, the package should
be examined using a checklist to verify that all the required elements are present.
4-2. Minimum Reporting Requirements.
a. As discussed in more detail below, each data package must include a cover page, a ta-
ble of contents, a Case Narrative, a Chain-Of-Custody form, a summary of the environmental
sample results (e.g., method of analysis, date analyzed, and amount analyzed), and a summary of
the batch QC results (e.g., method blank and LCS results). A summary of all instrument calibra-
tion results (i.e., initial calibrations, initial calibration verifications, and continuing calibration
verifications) and copies of the sample preparation, standard preparation, and instrument run log
sheets must also be included in the data package as described below.
b. The organization of the data package must be such that chemical data are reported on
a per batch basis. All calibration, method, and batch QC results must be presented on summary
forms using a tabular format. The use of CLP standard forms is not necessary. However, sub-
mission of standard instrument output alone is unacceptable to satisfy the reporting requirements
for PB data packages. Batch QC samples must be clearly linked to their associated environ-
mental samples. Instrument QC samples must be clearly linked to the associated environmental
and batch QC samples.
c. The data package must contain sufficient information to determine how the final sam-
ple concentrations were calculated from the calibration curves, or, alternatively, how a calibra-
tion standard may be expressed as a final sample concentration. In particular, using the sample
and calibration summary forms and the standard and sample preparation logs, it must be possible
to express the low calibration standard as a sample concentration.
4-2.1. Cover Letter.
The cover sheet includes the following information:
a. Title of Test Report or Test Certificate.
1
The evaluation of completeness performed during PB data review should not be confused with the
completeness evaluation that is performed during data usability assessment. The former relates to the data
package while the latter is more global in nature and is performed to determine whether or not there is sufficient
data of known and acceptable quality to support project decisions.
4-1
EM 200-1-10
30 Jun 05
b. Name and location of laboratory.
c. Laboratory point of contact with phone and facsimile numbers.
d. Name and location of any subcontractor laboratories.
e. Contract number.
f. Client name and address.
g. Project name and site location (if provided by client).
h. Statement of data authenticity and official signature and title of person authorizing the
release of the test report.
i. Amendments to previously released reports shall clearly identify previous reports and
state the reason(s) for the report amendments..
4-2.2. Case Narrative.
A Case Narrative must be included in each report. The Case Narrative contains tables summa-
rizing the samples received, providing a correlation between field sample numbers and labora-
tory sample numbers, and identifying which analytical test methods were performed. When
analyses are subcontracted to other laboratories, the Case Narrative must clearly specify which
laboratory performed each analysis. Samples that were received but not analyzed must also be
identified. Extractions or analyses that are performed out of holding times must be appropriately
noted. The Case Narrative must define all data qualifiers or flags used. Deviations of any cali-
bration standards or QC sample results from appropriate acceptance limits must be noted and as-
sociated corrective actions taken by the laboratory must be discussed. Any other factors that
could affect the sample results (e.g., air bubbles in VOC sample vials, excess headspace in soil
VOC containers, the presence of multiple phases, sample temperature and sample pH excursions,
container type or volume, etc.) must be noted.
4-2.3. Technical Summary.
Summary forms for each sample include the information specified below. Information need not
be repeated if noted elsewhere in the data package.
a. Laboratory name and location (city and state).
b. Project name and unique ID number (if specified by client).
c. Field sample ID number as written on custody form.
d. Laboratory sample ID number.
4-2
EM 200-1-10
30 Jun 05
e. Matrix (soil, water, oil, etc.).
f. Sample description.
g. Sample preservation or condition at receipt.
h. Date sample collected.
i. Date sample received.
j. Date sample extracted or prepared.
k. Date sample analyzed.
l. Analysis time when holding time limit is less than 48 hours.
m. Method (and SOP) numbers for all preparation, cleanup, and analysis procedures em-
ployed.
n. Preparation, analysis, and other batch numbers.
o. Analyte or parameter.
p. Method reporting limits adjusted for sample-specific factors (e.g., aliquot size,
dilution or concentration factors, and moisture content).
q. Method quantitation limits.
r. Method detection limits.
s. Final analytical results (e.g., concentrations) with the correct number of significant fig-
ures.
t. Data qualifiers and definitions of data qualifiers.
u. Concentration units.
w. Dilution factorsAll reported data shall reflect any dilutions or concentrations. The
dilution factor must be noted on the analytical report. If analyses were performed for both the
undiluted and diluted samples, both results must be reported.
x. Percent moisture or percent solids (e.g., soils, sediments, and sludges are typically re-
ported on a dry weight basis).
y. Sample aliquot analyzed.
4-3
EM 200-1-10
30 Jun 05
z. Final extract volume.
4-2.4. Sample Management Records.
These types of records include the documentation accompanying the samples (e.g., original
chain-of-custody record, shipping documents, and laboratory notification sheets), records gener-
ated by the laboratory which describe the condition of the samples upon receipt at the laboratory
(e.g., Sample Cooler Receipt forms, and any records of telephone conversations), and any re-
cords generated to document sample custody, transfer, analysis, and disposal.
4-2.5. Batch QC Summary Results.
a. The data package must include all batch QC sample results. This includes method
blank (MB), laboratory control sample (LCS), laboratory control sample duplicate (LCSD),
matrix spike (MS), matrix spike duplicate (MSD), matrix duplicate (MD), and post-digestion
spike (PDS) results with the associated acceptance criteria. Summary forms for the MS, MSD,
PDS, and LCS, must specify the spiking concentrations, the measured concentrations (e.g., the
measured concentrations before and after spike addition for MS and PDS analyses), the percent
recovery, and the percent recovery acceptance limits for each target analyte. The nature of the
recovery acceptance ranges must also be specified (e.g., project-specified acceptance range or in-
house laboratory statistical limits). The laboratorys statistical warning and control limits must
be specified (in addition to any project required acceptance ranges) for each target analyte for the
laboratory control samples (for each preparatory and determinative method).
b. Summary forms for replicate results (e.g., duplicate precision as measured by

MS/MSD, LCS/LCSD, or sample/MD pairs), must specify the concentrations of the replicate
results, the relative percent differences (RPDs) or percent relative standard deviations
(%RSDs) for each set of replicates (e.g., each duplicate pair), and the acceptance limits (e.g., for
the RPDs). The nature of the acceptance limits must also be specified (e.g., project-specified
limits or in-house statistical limits for the RPDs).
4-2.6. Standard Preparation Logs.
a. Copies of all relevant standard preparation log sheets must be provided for all
calibration standards and spiking standards associated with the environmental samples (e.g., the
initial calibration, initial calibration verification, continuing calibration verification standards as
well as the MS, PDS, and LCS spiking standards). At a minimum, the standard preparation logs
must clearly specify the following for all standards:
(1) Source (e.g., manufacturer and lot number for commercial stock solutions).
(2) Composition (e.g., the concentration of all target analytes, surrogates, and internal
standards).
(3) Date of preparation and expiration.
4-4
EM 200-1-10
30 Jun 05
(4) Name of the analyst.
(5) ID number of the standard.
(6) Reagents and solvents added to standards (including source and lot numbers).
b. When a standard is prepared via the dilution of a stock solution, the following must be
specified: The spiking volume and concentration of the stock solution, and the final volume and
concentration of the diluted standard. Copies of manufacturer certificates for commercially pur-
chased stock standards must be included in the standard preparation logs. When the laboratory
prepares its own stock solutions, calculations and conversion factors must be shown in the stan-
dard preparation log (e.g., using equations or sample calculations).
4-2.7. Sample Preparation Logs.
Copies of the sample preparation log sheets must be included in the data package. The sample
preparation logs must include the following information:
a. Sample and batch ID numbers.
b. Matrix.
c. Preparatory method (method and laboratory SOP ID number).
d. Date of sample preparation.
e. Initial volume or weight of the sample processed.
f. Final volume of sample processed (e.g., after digestion, extraction or cleanup).
g. Percent moisture (for solid samples).
h. Reagents and solvents added to the samples (including source and lot numbers).
i. Preservation and pH checks or adjustments.
j. Spiking standards (ID number of the LCS, PDS, and MS spiking solutions, volume
added, and the final spike concentration).
k. Name of the analyst.
4-2.8. Instrument Run-Sequence Logs.
a. Copies of instrument analysis log sheets must be provided for each instrument for
each day or analytical shift project samples or associated QC samples were analyzed.
Instrumental analysis logs are particularly important since they provide the basic link between
4-5
EM 200-1-10
30 Jun 05
the environmental sample analyses and QC data. The run sequence logs must include the
following information:
(1) Date of analysis.
(2) Determinative method (including SOP ID number).
(3) Name of the analyst.
(4) Unique ID numbers for environmental and QC samples.
(5) Amount of the sample (instrumentally) analyzed (e.g., the injection volume for
chromatographic methods).
(6) Instrument ID number and salient instrument features (column type for chroma-
tographic methods).
(7) Reanalyses and dilution factors (when performed).
b. The run log must clearly identify all QC samples (e.g., continuing calibration verifica-
tions). It must also clearly indicate which environmental and batch QC samples are associated
with each initial calibration, initial calibration verification (ICV), and continuing calibration
verification (CCV). The order in which environmental and QC samples (e.g., ICVs and CCVs)
are recorded in the run log must always be consistent with the temporal order in which the sam-
ples were actually analyzed. The time of analysis for CCVs and tunings (when performed) must
be specified for chromatographic methods. When an autosampler is used (i.e., any device that
enables the loading of multiple samples for sequential analysis) the position or sequence number
should be recorded in the run log (e.g., the purge port number for purge-and-trap analyses).
Lastly, any salient analytical problems (e.g., carry over) must be noted. In particular, when a QC
sample (e.g., a CCV) is reanalyzed, the run log must specify why the reanalysis was performed.
4-2.9. Traceability.
The data package must clearly demonstrate complete traceability of all standards. For example,
unique ID numbers must link all batch QC samples (e.g., MSs and LCSs) in the instrument run
log to the spiking solutions listed in the sample preparation log. The spiking solutions in the
sample preparation log must be traceable to the original (primary) stock standards via the stan-
dard preparation log. Similarly, unique ID numbers must link instrument QC samples (e.g.,
CCVs and ICVs) in the run log to the corresponding standards in the standard preparation log.
4-2.10. Calibration Summary Results.
a. The concentration and corresponding instrumental response (e.g., peak area or peak
height) must be reported for each initial calibration standard. When the initial calibration (and
quantitation) is performed using internal standards, instrumental response and the corresponding
internal standard concentration (or amount and volume of internal standard analyzed) must be
4-6
EM 200-1-10
30 Jun 05
reported for the initial calibration. When more than one internal standard is used, the data pack-
age must list the set of target analytes associated with each internal standard. A quantitative
goodness-of-fit value (e.g., the correlation coefficient, the coefficient of determination, or the
%RSD) must be reported for the calibration curve of each target analyte and surrogate (when
surrogates are used).
b. Plots of the initial calibration curves and instrumental printouts of quantitation sum-
mary reports must also be included in the data package for all target analytes. Instrument re-
sponse must be plotted on the y-axis as the dependent variable and concentration must be plotted
on the x-axis as the independent variable. The equation of each calibration curve must also be
specified. When calibrations are performed using response factors, the mean response factors
must be included with the data package (e.g., in lieu plots and equations for the initial calibra-
tions).
c. The true (i.e., reference) concentration (i.e., level spiked), measured concentration,
the instrumental response corresponding to the measured concentration, and percent recovery
must be reported for each CCV and ICV for each target analyte and surrogate (when surrogates
are analyzed). When internal standards are used, the instrumental response for each internal
standard and the corresponding internal standard concentration (or amount and volume of inter-
nal standard analyzed) must also be reported. When an ICV is not performed, the data package
must clearly indicate (e.g., in the standard preparation logs) whether an independent-source stan-
dard was used to spike the CCVs or LCSs.
4-2.11. Chromatographic Methods for Organic Target Analytes.
The additional reporting requirements specified in this section of the document apply only to
chromatographic methods for organic target analytes.
4-2.11.1. Initial Calibration.
When the initial calibration is performed using response factors, the response factor for each
initial calibration standard and the mean response factor for each analyte and surrogate must be
reported. The %RSD must also be reported for each set of initial calibration standards. A re-
porting format similar to the CLP Initial Calibration Summary Form for the GC/MS BNA and
VOA analyses is recommended. Response factors must also be reported for CCVs and the initial
calibration when minimum response factors are specified by the analytical method (e.g., GC/MS
Methods 8260A and 8270B). (Note that this does not imply that regression analysis cannot be
used to perform initial calibrations for these methods.)
4-2.11.2. Internal Standard Summary Information.
When internal standards are used, the internal standard areas and retention times must be sum-
marized for each batch QC and environmental sample in a tabular format (e.g., using summary
forms similar to the CLP Internal Standard Area And RT Summary forms). The internal stan-
dard retention time and retention time windows must also be specified for the most recent asso-
ciated CCV. The area counts and area count acceptance ranges for the associated mid-level ini-
4-7
EM 200-1-10
30 Jun 05
tial calibration standard must be specified on the internal standard retention time and peak area
summary form.
4-2.11.3. Surrogate Results.
The expected concentrations of the surrogates, the surrogate recoveries, and surrogate recovery
acceptance ranges must be reported for all the environmental samples, batch QC samples, con-
tinuing calibration verifications, and initial calibration verifications. The surrogate recoveries
for the environmental and batch QC samples must be summarized in a tabular format (e.g., using
forms similar to the CLP Surrogate Recovery forms). The nature of the surrogate recovery
acceptance ranges must also be specified (e.g., project-specified versus in-house statistical con-
trol ranges). The laboratorys statistical warning and control limits must be specified (in addi-
tion to any project required acceptance range) for the surrogates in the laboratory control sam-
ples or method blanks (for each preparatory and determinative method).
4-2.11.4. Chromatographic Methods With 2-D Detectors.
a. Additional reporting requirements are required for chromatographic methods with 2-D
detectors (e.g., FIDs, PIDs, and ECDs). When chromatographic analyses are performed using
second column confirmation, the results for both analytical columns must be reported (e.g., sur-
rogate recoveries, sample results, and initial and continuing calibration results).
b. The results for the environmental samples must be reported using summary forms
similar to the CLP Pesticide Identification Summary For Single Component Analytes and the
Pesticide Identification Summary For Multicomponent Analytes. Retention times and reten-
tion time windows must be reported for all single component standards (e.g., initial calibration,
continuing calibration verification, and internal standards), target analytes, and surrogates or
both the primary and confirmatory columns. For multi component analytes, the retention times
and retention time windows should be specified for at least three to five characteristic peaks.
(Note that retention time and retention time windows must be reported for both the primary and
confirmatory columns.) In addition, the results (e.g., concentrations) from the primary and sec-
ondary columns, as well as the corresponding RPDs must be reported.
4-3. Evaluation of Completeness.
a. Use professional judgement to determine the degree to which missing information can
be tolerated. Distinguish sporadic occurrences of missing noncritical data from systematic non-
compliances. For example, if the LCS and method blank results were not included, the data
would typically be of unknown quality and would be rejected. However, the entire data package
would not typically be rejected if the dilution factor were not specified for one of the environ-
mental samples.
b. If the data package is not substantively complete, (i) missing information must be re-
quested from the laboratory, (ii) the data package must be rejected, or (iii) a limited data review
must be performed. When the data package is grossly deficient, the reviewer should consult with
the Project Manager to determine which option is most appropriate for the data objectives of the
4-8
EM 200-1-10
30 Jun 05
investigation. For example, if missing deliverables cannot be obtained from the laboratory or if
data evaluation cannot be delayed because of scheduling constraints (e.g., because time-critical
decisions must be made), a more limited data review may be appropriate. If the data are being
used to support critical decisions (e.g., for litigation), it may be necessary to reject the data when
standards are not traceable or the COC is missing (since the integrity the data has not been de-
finitively demonstrated).
c. When QC sample results are not included in the data package, it may be possible to
use the available results to evaluate or to make inferences concerning the quality of the data.
However, this approach must be used with caution since the missing QC samples may have been
analyzed by the laboratory but not reported because of QC failure. Some evaluation strategies
for incomplete data are discussed below.
4-3.1. Missing Blanks.
a. When a blank is missing, a higher hierarchy blank may be used to qualify the corre-
sponding field samples for contamination. For example, if the method blank is not available
(e.g., because it was not processed with the batch of samples), then field samples may be quali-
fied using another associated blank that was processed with the batch of samples. In particular,
if a rinsate blank or trip blank were analyzed with the batch of samples, rather than rejecting the
data, the data could be evaluated using the trip blank or rinsate blank data. The trip blank or rin-
sate blank would be indicative of the accumulative field and laboratory contamination.
Note: This approach will not be appropriate when the field blank is not processed in the
same batch (or in the same manner) as the environmental samples for which it is as-
sumed to represent.
b. Similarly, an environmental sample in the preparation batch for which no target ana-
lytes were detected may serve as a field blank (since such a sample would demonstrate the lack
of systematic field and laboratory contamination).
Note: Since samples are processed through a variety of handling, preparatory, and
analysis procedures, blanks are typically collected during various stages of these proce-
dures in a manner which would establish the source of contamination and enable the
implementation of corrective action. In theory, if a comprehensive blank is truly rep-
resentative of all contamination that could have been introduced from the time of sam-
ple collection to analysis (especially if the blank is free of contamination), then missing
lower hierarchy blanks should not affect the data qualification. However, it should be
noted that, in practice, it is difficult to obtain a representative comprehensive blank.
For example, common laboratory contaminants such as methylene chloride can appear
in blanks in a sporadic manner (e.g., may be present in a method blank but may not be
present in a higher hierarchy blank such as a field blank).
c. When the method blank is missing and a higher hierarchy blank is not available to
evaluate the data, qualification may be required for the corresponding field sample results.
(Contractual corrective actions may also be required for missing method blanks, because labo-
4-9
EM 200-1-10
30 Jun 05
ratories are usually required to process a method blank with each batch of field samples.) In
general, nondetections must not be qualified (unless other QC problems are present). Detec-
tions must be qualified with the R, X, or N flag. Project-specific DQOs and professional judg-
ment must be used to determine which flag is more appropriate. However, the R or X flag must
be used to qualify detections in the absence of a technically defensible rationale for using the N
flag. Some situations for which the use of the N flag may be appropriate are presented below.
d. It may be appropriate to qualify an analyte detection with the N flag if the analyte was
reliably detected in other site field samples that were processed in a separate batch with an ac-
ceptable method blank. Since the probability of external contamination as the source of a detec-
tion typically varies inversely with the magnitude of the detected concentration, it may be appro-
priate to qualify very high level detections with the N flag (e.g., detections ten times greater than
the MQL). If an action level is available, it may be appropriate to qualify detections less than the
action level (especially very low-level detections) with the N flag. However, in the absence of a
technically defensible rationale to do otherwise (i.e., when information supporting the validity of
detected concentrations is not available), detections greater than an action level must be qualified
with the R or X flag when blank results are not available.
4-3.2. Missing Laboratory Control Samples.
In general, when LCS results are not available for a particular batch of field samples, the data are
of unknown quality and the associated field sample results must be qualified with the R or X
flag. Since a laboratory is typically required to process at least one LCS with each batch of sam-
ples, contractual corrective action for unacceptable performance may be appropriate. Possible
exceptions are discussed below.
4-3.2.1. Matrix Spike Data.
a. If the LCS is missing but all target analytes are present in the MS and precision
information is available from the MS/MSD or MD, then the associated field samples may be
qualified (for bias and precision) using the MS/MSD or MS/MD results since this data gives bias
and precision information for the overall method. In particular, if all the MS/MSD or MS/MD
recoveries and RPDs are in control, then the sample results need not be qualified. When the
MS/MSD or MS/MD is not in control, the associated field samples must be qualified but the QC
failure may be due to poor laboratory method performance (rather than matrix interference).
b. If the LCS is missing and only a subset of the target analytes is present in a MS/MSD
or MS/MD pair, then qualify the associated field samples as discussed above for the subset of
spiked analytes. Reject or tentatively reject the associated field sample results for the unspiked
target analytes. If the data are being used to support noncritical decisions, (for target analytes
not present in the matrix spike), it might be appropriate to reject or tentatively reject nondetec-
tions but to qualify detections as estimated.
4-10
EM 200-1-10
30 Jun 05
4-3.2.2. Surrogate Spike Data.
Surrogate recoveries may be used to make inferences about the performance of organic analyses.
If acceptable surrogate results are available and the data will be used to support noncritical deci-
sions, it is recommended that the data be qualified as estimated rather than as rejected. However,
this approach must be used with caution since the surrogates may not be representative of all the
target analytes (e.g., although ketones are often target analytes for SW-846 Method 8260B, none
of the surrogates recommended in the method are ketones).
4-3.2.3. CCV Data.
When all environmental and QC samples undergo the same preparatory and determinative proc-
esses or when significant sample preparation is not performed (e.g., aqueous VOCs by purge-
and-trap GC/MS and aqueous anions by ion chromatography), CCVs are essentially LCSs.
Method performance may be evaluated using the CCV results. If the CCV results are acceptable,
then the data would not be qualified on the basis of the missing LCS results.
4-3.3. Missing Detection Limits.
a. Detection limits must be included in the data package. Laboratory detection limits are
typically the method detection limits (MDLs) defined in 40 CFR, Part 136, Appendix B. In-
strument detections limits (IDLs) must not be viewed as a substitute for MDLs. For example,
MDLs may be significantly higher than IDLs when an analytical method involves extensive sam-
ple preparatory procedures.
Note: Laboratories frequently perform an MDL study for a single instrument and erro-
neously report the resulting MDLs as applicable to all similar instruments used for the
same analytical method. Method detection limits are matrix, method, and instrument-
specific. For example, if a laboratory performs Method 8260B using six GC/MS in-
struments, then six sets of MDL studies must be available or sensitivity must otherwise
be demonstrated for all six instruments. A valid conservative approach may consist of
performing a separate MDL study for each instrument but reporting only the highest
MDLs for each instrument.
b. When detection limits are not available (i.e., not included in the data package), the
laboratory should be contacted for this information or the method reporting limits (MRLs) for
nondetections must be set equal to the MQLs (e.g., when the laboratory lists reporting limits that
are less than the MQLs). However, the MQLs must be verified as discussed in Chapter 6.
(1) If an action level is unavailable, then qualify all detections less than the MQL with
the J flag. Report all nondetections as < MQL or MQL U, where MQL denotes the
numerical value of the method quantitation limit.
(2) If an action level is available, then compare the MQL to the action level. If the MQL
is less than the action level, then qualify detections and nondetections as described above. How-
4-11
EM 200-1-10
30 Jun 05
ever, if the MQL is greater than the action level, adequate sensitivity has not been demonstrated.
Qualify nondetections with the X or XU flag and qualify detections less than the action level with
the X flag. A detection below the action level does not demonstrate that the target analyte is ac-
tually present in the environmental sample below the action level because analytical uncertainty
is large or undefined below the quantitation limit. Detections greater than the action level but
less than the MQL may be qualified with the X flag but, at a minimum, must be qualified with
the J flag. When resampling or reanalysis cannot be performed, the use of the J flag normally
constitutes the more conservative approach.
4-12
EM 200-1-10
30 Jun 05
CHAPTER 5
Holding Times and Preservation
5-1. Introduction.
The primary objective of this review is to ascertain the representativeness of the analytical data
in the context of preservation and holding time limits1. Holding times for environmental sam-
ples are calculated from the dates of sample collection to preparation and analysis. (Refer to the
glossary).
5-2. Acceptance Criteria.
a. The data package must clearly indicate the dates of all sample handling processes and
the method of sample preservation. Holding time and preservation acceptance criteria and cor-
rective actions are determined by method requirements and project DQOs. If holding time or
preservation requirements are not specified for the project, use the published holding times and
preservation requirements that are summarized in Table B-1 (Appendix B).
Note: The preservation requirements and holding time limits were primarily taken from
SW-846 and water methods (40 CFR, Part 136.3). Other sample preservation and
holding time criteria may be more applicable (e.g., depending on the methods selected
and matrices being tested).
b. Published holding time limits are generally considered maximum times that samples
may be held before analysis and still be considered compliant with method guidelines, and typi-
cally apply to preserved samples.
Note: Published holding times listed in environmental methods and regulations are not
necessarily scientifically valid. However, use of alternative holding times may have a
profound impact on the legal defensibility of results. It is recommended that published
holding times be extended or shortened, if the client and regulators have agreed to other
holding times for a particular project (e.g., based upon the chemistry of the method and
holding time studies).
5-3. Evaluation.
a. Holding times and preservation are evaluated using the COC (Chain Of Custody)
form, the laboratorys Cooler Receipt form (e.g., refer to the format of the USACE Cooler
Receipt form), Case Narrative, sample preparation logs, and instrument run logs.
1
Sampling design (e.g., sampling locations) probably affects representativeness far more than any other factor. In
the context of a full data usability assessment, the review described here should be viewed as a screening process to
determine if the samples are potentially representative of the environmental matrices being sampled.
5-1
EM 200-1-10
30 Jun 05
(1) Check sample holding times using the date of sample collection listed on the COC
form with the dates of sample preparation and analysis reported on the laboratorys sample
summary forms.
(2) Verify that digestion, extraction, or cleanup dates reported on the sample summary
forms are identical to the dates listed on the sample preparation log sheets.
(3) Verify that the analysis dates reported on the sample summary forms are identical to
those listed on the instrument run logs.
(4) Review the Case Narrative and Cooler Receipt form included in the data package to
determine if all the samples were properly preserved and holding times were met. Note any
problems that may have impacted the integrity of the samples (e.g., samples not maintained at
2oC - 6oC, aqueous VOC samples with head space, custody seals that are broken, and holding
times that are not met). Verify that the pH of chemically preserved samples was checked and
appropriate pH values were obtained.
b. If samples are properly preserved, holding time limits are met, and no problems with
the samples are indicated in the Case Narrative, the laboratorys Cooler Receipt form, and the
COC form, then assume that the physical integrity of the samples is acceptable.
c. If there are holding time violations or preservation problems, then the integrity of the
samples may have been compromised. In the absence of information to the contrary, assume that
holding time and preservation problems give rise to a low bias.
Note: This assumption typically constitutes a conservative approach but will not always
be appropriate. For example, the assumption will not be valid when degradation prod-
ucts are also target analytes. In particular, holding time noncompliances would pre-
sumably give rise to a low bias for pesticides such as DDT and Endrin, but to a high
bias for associated degradation products such as DDE and Endrin aldehyde. In addi-
tion, when samples are held for an extended period of time, target analyte may leach or
permeate into the storage containers, giving rise to a high bias. To address this poten-
tial problem, the representativeness of the method blanks would need to be evaluated.
For example, if the method blank and samples were stored and analyzed together after
the holding time limit, the absence of blank contamination in the method blank would
suggest that leaching and permeation did not give rise to a high bias.
d. Distinguish gross holding time and preservation noncompliances from marginal

noncompliances. Using the guidance presented in the USEPA Contract Laboratory Program
National Functional Guidelines For Organic Data Review, any holding time that is greater than
twice the holding time limit is considered to be a gross holding time noncompliance. However,
if holding time studies were performed, a gross holding time noncompliance would be defined
on the basis of these studies.
5-2
EM 200-1-10
30 Jun 05
e. Professional judgment is critical for evaluating holding time noncompliances. The

evaluation of preservation problems is highly dependent upon the nature of the target analyte, the
matrix, the method of analysis, and the sample handling procedures. For example, if aqueous
samples for alkalinity were chemically preserved using pH adjustment, the analyses would be
considered unusable. However, BNA soil sample results would not be rejected if a cooler tem-
perature of 7C were reported. Because of the time required to reach thermal equilibrium (rela-
tive to that required to ship the samples), samples may not cool during shipment to the 26C
acceptance range even when adequate refrigerant is placed in the cooler.
5-4. Qualification.
The data qualification strategies presented in this section of the document are conservative in
nature. A holding time noncompliance is assumed to give rise to a low bias.
5-4.1. Low Stability Target Analytes.
If there is a holding time noncompliance or a preservation problem for analytes that are known to
readily volatilize or degrade in the matrix being tested (e.g., aqueous aromatic VOCs and
hexavalent chromium), then qualify the results as follows:
a. Qualify all nondetections with the R flag. For example, if large bubbles are reported
in vials of the aqueous VOC samples or unpreserved aromatic VOC samples were analyzed
beyond the limit, then qualify all nondetections with the R flag.
b. At a minimum, qualify all detections with the J- flag. If an action level is specified
and the analyte is detected at a concentration less than the action level, then qualify the detection
with the X flag. Although the detection may be qualitatively reliable, it is not quantitatively reli-
able. Although the analyte was detected (despite the noncompliant holding time or preservation
problem), the reported result potentially possesses a low bias and does not demonstrate that the
analyte is actually present in the environmental sample at a concentration less than the action
level.
5-4.2. High Stability Target Analytes
If there is a holding time or preservation noncompliance for analytes that are relatively stable in
the matrix being tested (e.g., dioxins and trace metals such as lead in soils), then qualify the data
as follows:
a. If the holding time limit is marginally exceeded or the preservation problem does not
appear to be significant, then qualify nondetections with the UN flag and detections with the J-
flag.
b. If there is a gross holding time or significant preservation problem, then qualify the
data as discussed in Paragraph 5-4.1; namely, qualify nondetections with the R flag and detec-
tions with the J- flag or X flag.
5-3
EM 200-1-10
30 Jun 05
Table 5-1
Data Qualification for Holding Time Noncompliance
Stability Holding Time (t) 1 Flag Remarks 2

Low t HTL None Holding time limit is met.
R y < MRL
X MRL < y < AL
t > HTL J- y > MRL and y > AL
High t HTL None Holding time limit is met.
UN y < MRL
HTL < t 2 HTL J- y > MRL
R y < MRL
X MRL < y < AL
t > 2 HTL J- y > MRL and y > AL
Notes: 1. The project-required holding time limit and the calculated holding time for the sample are denoted by HTL
and t, respectively. 2. The concentration of the field sample, the action limit, and the method reporting limit are
denoted by y, AL, and MRL, respectively.
5-4
EM 200-1-10
30 Jun 05
CHAPTER 6
Data Review Reports
6-1. Introduction.
The primary objective of this review is to ensure that analytical sensitivity is adequate for pro-
ject-specific action levels and to ensure that data are reported in a manner that is consistent with
the laboratorys detection and quantitation limits. The evaluation of sensitivity will be a function
of how the detection, quantitation, and reporting limits are defined and whether or not action
levels are specified. Since it is impractical to discuss sensitivity in the context of multiple defi-
nitions for these limits, they will be defined as the method detection limits (MDLs), method
quantitation limits (MQLs), and method reporting limits (MRLs) presented the glossary. In
particular, it is assumed that the detection limit is the method detection limit of 40 CFR, Appen-
dix B, Part 136. The method quantitation limit is defined (primarily) as the low level calibration
standard adjusted for method specific factors. Lastly, the method reporting limit is defined as
the threshold or censoring limit below which target analyte concentrations are reported as ND
(i.e., not detected) or as < (i.e., less than).
6-2. Method Reporting Limits.
6-2.1. Establishing Method Reporting Limits.
a. The definition of the MRLs must be declared in each data package or in project docu-
ments such as the Quality Assurance Project Plan (QAPP).
Note: Merely listing numerical values for the MRLs will not satisfy this reporting re-
quirement; the MRLs must be defined in terms of the laboratorys actual quantitation
and detection limits.
b. In general, any analyte concentration greater than the detection limit may potentially
be reported either as a detection or as a nondetection with respect to some censoring
(reporting) limit greater than the detection limit. For example, if an action level, AL, is very
large relative to the MQL, then it may be desirable to report all analyte concentrations less than
5% of the AL action level as < MRL or MRL U, where MQL < MRL = 0.05 AL. In this
context, < MRL indicates that (i) the analyte is present below the detection limit, or (ii) was
detected at some concentration greater than the detection limit but less than 0.05 AL.
Conversely, if low-level reporting is desirable, then it may be appropriate to establish a
censoring limit (MRL) at some concentration greater than the method detection limit (MDL) but
less than the MQL. Under these circumstances (MDL < MRL < MQL), analyte concentrations
between the MRL and MQL would be reported as estimated and concentrations less than the
MRL would be reported as < MRL.
Note: The term reporting limit is being defined in a more general manner than is con-
ventionally used for environmental testing. For example, according to the CLP State-
ment of Work (SOW) for organic analyses, the reporting limit for nondetections is nec-
6-1
EM 200-1-10
30 Jun 05
essarily the CRQL. If an analyte is not detected, the reporting limit is the CRQL and
detections below the CRQL are reported as estimated. However, there is no a priori
reason for setting the reporting limit equal to the quantitation for all data uses. For
example, presence-absence issues can typically be resolved at concentrations that are
significantly less than the quantitation limits.
c. In the absence of project-specific guidance, assume that all reliable detections greater
than the MDL or MRL must be reported. (Note that detections should be reported based upon
the laboratorys detection limits as well as the analysts judgement.) In addition, assume that the
MRL for a nondetection must be no less than the limit of identification (LOI) or the reliable
detection limit (RDL). The RDL and LOI are approximately two times the MDL.
Note: Establishing any reporting limit constitutes a form of data censoring. Censor-
ing results (i.e., reporting nondetections to twice the MDL) will typically be appropri-
ate when action levels have been established and the results (detections and nondetec-
tions) are being compared to the action levels on a point-by-point basis. However, this
approach will not be appropriate for all projects. For statistical applications, it is usu-
ally desirable to report results without any censoring (e.g., to report results less than the
MDL). The reviewer must refer to project-specific objectives prior to performing cen-
soring or evaluating the data with respect to the reporting limits.
6-2.2. Qualification.
a. If the reporting limit is less than the LOI or the RDL (i.e., two times the MDL), then
qualify nondetections (at the reporting limit) with the UN flag and discuss the potential high
false negative probability at the reporting limit in the data review report. Alternatively, if the
project action levels (ALs) are relatively high (e.g., at least 10 to 20 times greater than the
MQL), increase the reporting limit to the quantitation limit (if the quantitation limit was estab-
lished by the lowest calibration standard) and qualify nondetections with the U flag.
b. If an action level (AL) is available, compare the MRL to the AL. If the MRL is
greater than the AL, qualify nondetections with the X or XU flag (since false negatives have not
been adequately addressed).
c. It is recommended that the MRL be no higher than %5 to 10% of the AL. If the MRL
is less than but near the AL, then use professional judgement to qualify nondetections, especially
when the AL is less than the MQL or the LCS acceptance limits are wide. For example, if the
MQL = 50 ppb, the MRL = 10 ppb, AL = 15 ppb, the LCS acceptance range is 50150% (e.g.,
for a 100 ppb spike near the mid-calibration range), and the LCS recovery associated with a set
of environmental samples is 55%, then nondetections reported as < 10 ppb do not demonstrate
the 15-ppb action level was met. Under these circumstances, nondetections would be qualified
with the X or XU flag
6-2
EM 200-1-10
30 Jun 05
6-3. Method Quantitation Limits.
6-3.1. Establishing Method Quantitation Limits.
a. Project planning documents (e.g., the QAPP) must define what constitutes a quantita-
tion limit. In general, project documents should specify tolerances for uncertainty at the quanti-
tation limit and strategies for verifying the tolerances have been satisfied (e.g., a low-level LCS
at the quantitation limit must be recovered to within 20% of its expected value). Unfortunately,
quantitation limits are often poorly defined.
Note: The laboratorys reported MQLs must not be evaluated solely upon the basis of
Practical Quantitation Limits (PQLs) or Contract Required Quantitation Limits
(CRQLs) specified in published analytical methods or project documents unless these
quantities are adequately defined (e.g., tolerances for uncertainty at the quantitation
limits are specified).
b. The guidance presented below will typically be applicable.
(1) A low-level LCS or CCV (spiked with the target analytes at or near the MQL) may
have been analyzed to verify the quantitation limit. Low-level CCVs would be appropriate for
methods that do involve significant sample preparation or for methods in which the calibration
standards are prepared with the environmental samples. Low-level CCVs can often be used to
verify the quantitation limits for inorganic methods (e.g., when the sample preparatory process
does not introduce too much uncertainty). However, this approach will not be valid for methods
that involve significant sample preparation and the CCVs are not processed with the environ-
mental samples. Under these circumstances, a low-level LCS (spiked with target analytes at or
near the MQL) is required to verify the quantitation limit.
(2) If a low-level CCV (e.g., the lowest calibration standard) was used to check the
MQL, verify that the CCV was recovered to within the tolerance for instrumental uncertainty
(the acceptance limits must be equal to or slightly greater than the acceptance limits for mid-
level CCVs). For example, for trace metals by ICP, the low-level CCV should be recovered to
within 10% to 15% of its expected value. If a low-level LCS was used to check the MQL, then
verify that the low-level LCS was acceptably recovered.
(3) If a low-level CCV or LCS spiked at the MQL or near the MQL (e.g., less than two
times the MQL) was not analyzed, then compare the reported MDL to each corresponding MQL
as discussed below (i.e., verify that each MQL is at least five to ten times greater than the MDL
and was established using the lowest initial calibration standard).
(4) Use the calibration data to verify that the laboratorys reported quantitation limit for
each analyte is established from the lowest calibration standard (or corresponds to a higher con-
centration that is within the calibration range).
6-3
EM 200-1-10
30 Jun 05
Note: This is not a sufficient condition to verify the project-required method quantita-
tion limits. It is often erroneously concluded that if the initial calibration curve is ac-
ceptable (e.g., as indicated by a high correlation coefficient), then the lowest calibration
standard will be acceptable for establishing the MQL. However, an acceptable fit for
the entire calibration curve does not necessarily imply that the uncertainty will be ac-
ceptable at concentrations near the lowest calibration standard. Conventional measures
of fit are not adequately sensitive to high variability at the low concentration ranges.
For example, when regression analysis is used to fit initial calibration results, a high
correlation coefficient is possible when the lowest standard radically deviates from a
linear fit (e.g., when instrumental response is inherently nonlinear at low concentra-
tions).
(5) Compare the MDLs (if available) to the corresponding MQLs to ensure that the
quantitation limits are sufficiently greater than the detections limits. If the MQL is established
from the lowest calibration standard but is not otherwise defined, ensure that the MQL is at least
five to ten times greater than the method detection limit.
Note: The quantitation limit will be dependent upon the magnitude of the analytical
noise (whether chemical or electronic in nature) that constitutes the background sig-
nal or response for the analysis method, and the project-required tolerance for uncer-
tainty for quantitation. Since the detection limit is measure of background response,
the quantitation limit must typically be greater than the detection limit by some multi-
plicative factor in order to meet the project-required error tolerance. In general, when a
low error tolerance is required, the quantitation limit must be significantly greater than
the detection limit.
If it is assumed that the magnitude of the analytical uncertainty is approximately

MDL, then the relative uncertainty will be about 20% at five times the MDL and
10% at ten times the MDL. (It is being assumed that the standard deviation determined
from the MDL study is not strongly dependent upon concentration and there is no sig-
nificant bias.) However, the actual relative uncertainty will often be higher than 10%
to 20% at five to ten times the MDL (e.g., because the standard deviation is often an in-
creasing function of concentration).
(6) If the laboratorys reported quantitation limit is less than the method quantitation
limit calculated from the lowest initial calibration standard and the standard is at least five times
greater than the MDL, then increase the quantitation limit using the lowest calibration standard.
(7) If the lowest calibration standard is not at least five times greater than the MDL and
an acceptable low-level CCV or LCS was not analyzed to verify the MQL, then the initial
calibration results must be evaluated. If the low-level calibration standard is less than five times
the MDL, it may be appropriate to use the next highest calibration standard to establish the
MQL. If possible, use the equation for the initial calibration curve to calculate the concentration
of the lowest calibration standard (i.e., calculate the concentration of the lowest standard from
the measured response) and ensure that the calculated value of the lowest standard is within the
uncertainty tolerance for the CCV. If it is not possible or practical to determine the MQL from
6-4
EM 200-1-10
30 Jun 05
the calibration data, then set the MQL to five to ten times the MDL, but indicate that the MQL is
an estimate in the data evaluation report. Multiply the MDL by at least a factor of ten for ICP
analyses.
6-3.2. Qualification.
a. Once the MQLs have been verified or established, qualify all detections less than the
MQLs as estimated using the J-flag (e.g., unless the X or R flag is more appropriate because sig-
nificant QC problems are observed).
b. If action levels are available, compare the MQLs to the action levels and ensure that
the MQLs are less than the action levels. Although the MQLs should have been compared to the
projects action levels during the planning stages of the project, sensitivity problems may still
occur (e.g., because of dilutions). As a rule of thumb the MQL should not be greater than
about one half of the AL for inorganic analyses and about one third of the AL for organic analy-
ses.
c. If the MQL is greater than a corresponding AL, adequate sensitivity has not been
demonstrated; qualify detections less than the AL with the X flag. Under these circumstances
(MQL < AL), depending upon project DQOs, it may be appropriate to also qualify detections
greater than the AL with the X flag (e.g., when a conservative estimate of contamination is not
desirable).
Table 6-1
Data Qualification for Sensitivity When Action Levels Are Available
Sample Result (y) Flag Remarks

LOI MRLs 1
y < MRL < AL U
y < MRL, MRL > AL X, XU Nondetections
MRL y < AL < MQL X
MRL AL < y < MQL J or X 2
MRL y < MQL < AL J Detections

MRL MQL < y No flag
Notes: 1. The action level, method reporting limit, and method quantitation limit are denoted as AL, MRL, and
MQL, respectively. The concentration of the target analyte in a field sample is denoted as y. (It is assumed that the
limit of identification is less than or equal to the MRL.) 2. A detection above the AL was obtained. However,
because quantitative uncertainty is high, the target analyte may not actually be present in the sample at a
concentration that exceeds the AL; the X flag may be appropriate. The use of the J flag constitutes a conservative
interpretation of the data (namely, that the AL has been exceeded).
6-5
EM 200-1-10
30 Jun 05
CHAPTER 7
Initial Calibration
7-1. Introduction.
The initial calibration is evaluated to ensure that the instrument was capable of producing ac-
ceptable quantitative data prior to the analysis of environmental samples. The concentration
range and number of standards for the initial calibration will be dependent upon the instrument,
method, and objectives of the project. The variation in instrumental response with concentration
may define a straight line or some curve. Instrumental response may be expressed either as peak
area (e.g., determined from a sum of detector signals) or as peak height (e.g., the maximum de-
tector signal above background noise).
7-2. Acceptance Criteria.
The calibration criteria presented below are based upon the guidance presented in USACE Shell
document and SW-846 (Method 8000B). However, the criteria are conservative and should be
applicable to most instrumental methods. Calibration may be performed using linear or nonlin-
ear fits. However, linear calibrations should be used in preference to nonlinear calibrations
7-2.1. Frequency.
a. An initial calibration must be performed prior to the analysis of samples and when a
continuing calibration verification is unacceptable.
b. For inorganic analyses, the initial calibration is typically performed at the beginning
of each analytical shift in which analyses are performed; that is, each time the instrument is set
up to perform analyses (e.g., turned on and warmed up). When analyses are performed con-
tinually, the initial calibration is typically performed on a daily basis (i.e., every 24 hours).
7-2.2. Number of Calibration Standards.
a. The number of calibration standards (or points) is highly method dependent. The
number of calibration standards will be proportional to the variability of instrumental response.
For example, the higher the variability of response, the greater the number of calibration stan-
dards that will be required. As a general rule, at least three calibration standards should be used
for linear calibrations. Additional calibration standards should be used if the linear calibration
range is greater than one or two orders of magnitude or when nonlinear calibrations are per-
formed.
b. When calibration curves are fitted with polynomials using regression analysis, the
number of initial calibration standards must be sufficient for at least one statistical degree of
freedom. The degrees of freedom for a regression curve, df = n - 1 - k, where k > 0. The variable
k is the order of the polynomial and n is the number of initial calibration standards. For example,
if a regression line is being used (k = 1), then the number of calibration standards (n) must be
7-1
EM 200-1-10
30 Jun 05
greater than or equal to three. However, beyond the minimum number of standards required to
perform the actual mathematical fit (e.g., three standards for linear regression lines), the number
of initial calibration standards required is somewhat arbitrary: Any number standards may be
used to generate a calibration curve that meets the specified tolerance for uncertainty. However,
when regression analysis is not being performed, only a single calibration standard may be re-
quired. For example, if response has been demonstrated to be linear through the origin (e.g., as
in ICP analyses), the initial calibration may be performed using only a single high-level standard.
c. It is recommended that the number of standards be equal to or greater than that speci-
fied in the most applicable promulgated USEPA method. For example, if SW-846 methods are
specified in the Sampling and Analysis Plan (SAP), then the guidance presented in the individual
SW-846 method should be followed to the extent that is appropriate for the objectives of the
project. For the purposes of illustration, the guidance in Update III of SW-846 for linear cali-
brations is summarized below. However, additional criteria are also specified.
(1) At least five calibration standards are used for all single-component target analytes
and surrogates. (Single-standard calibrations are not typically acceptable for surrogates).
(2) A blank and a minimum of three calibration standards are used for SW-846 trace
metal analyses (e.g., GF-AA, FL-AA, and ICP analyses). However, a blank and a minimum of
five calibration standards are used for mercury analyses.
(3) A blank and one calibration standard may be used for ICP analyses. However,
though not specified in Method 6010B, the initial calibration line should be initially verified
using a low-level CCV in order to establish the method quantitation limit.
(4) Calibration for multi component analytes such as Aroclors are performed using a sin-
gle standard. Instrumental response is related to concentration using the peak areas or heights of
several peaks (e.g., at least three characteristic peaks for the Aroclors). However, contrary to
the guidance presented in the method, a minimum of three calibration standards is recommended
for chromatographic analyses for multi-component analytes.
7-2.3. Linear Calibration.
Linear calibrations must be performed using regression analyses (with the possible exception of
chromatographic analyses). Instrumental response, as measured by the slope of the calibration
line, must be high relative to analytical uncertainty (e.g., calibration lines with very shallow
slopes would not acceptable).
7-2.3.1. Organic Methods.
The coefficient of determination (i.e., the square of the correlation coefficient r) must be equal to
or greater than 0.980 for each target analyte. Mean response factors may be used to perform lin-
ear calibrations through the origin for chromatographic analyses. The percent relative stan-
dard deviation (%RSD) for the mean response factors must be equal to or less than 15% for
each target analyte. However, if only three calibration standards are used (e.g., for multi compo-
7-2
EM 200-1-10
30 Jun 05
nent analytes such as PCBs), the coefficient of determination should be 0.990 or greater or the
%RSD should be 10% or less. In addition, any method-specified minimum response factors
must be met.
Note: A number of analytical methods (e.g., SW-846 methods) specify a maximum

%RSD of 20% for the initial calibration for chromatographic methods with 2-D detec-
tors. An acceptance limit of 20% is not recommended. Initial calibration lines with
%RSDs of 20% often exhibit poor linear fits (e.g., a calibration line may not adequately
fit the data at the upper end of the calibration range or the regression coefficient may be
less than 0.99).
7-2.3.2. Inorganic Methods.
The coefficient of determination (square of the correlation coefficient r) must be at least 0.990
for each target analyte. Note that when a single standard is used to perform the initial calibration
for ICP analyses, a correlation coefficient cannot be calculated. Under these circumstances, the
initial calibration must be evaluated using CCVs at multiple concentrations.
7.2.4. Nonlinear Calibration.
a. Nonlinear calibrations are appropriate when linear calibrations cannot be performed

over a sufficiently wide working range (e.g., when detector response is inherently nonlinear over
a calibration range that spans two orders of magnitude or less). Nonlinear calibrations are inap-
propriate to compensate for detector saturation at higher concentrations or to avoid proper in-
strument maintenance. A large reduction in instrumental response (e.g., curve flattening char-
acteristic of detector saturation) must not occur near the upper portion of the curve correspond-
ing to high concentrations. Instrumental response must be high relative to analytical uncertainty
and must be similar in magnitude to that for linear calibration.
b. All nonlinear calibrations must be performed using regression analysis. Nonlinear

calibration curves must be generated using polynomial fits of no higher than third order (e.g., y =
ax3 + bx2 + cx + d) and must possess at least three statistical degrees of freedom. The coefficient
of determination must be equal to or greater than 0.99.
7-2.5. Intercept of Calibration Curve.
A calibration curve (whether linear or nonlinear) must not be forced through the origin unless it
is demonstrated (e.g., during method development) that the intercept (i.e., y[x = 0]) is not statis-
tically different from zero (e.g., by performing a t-test for the y-intercept or comparing it to the
MDL.) Arbitrarily forcing a calibration curve through the origin may adversely impact low-level
quantitative results. Similarly, when calibration curves are generated using regression analysis,
the curves must not be artificially weighted toward the origin by including the point (0,0) one or
more times.
7-3
EM 200-1-10
30 Jun 05
7-3. Evaluation.
Review the standard preparation and run sequence log sheets to verify that the initial calibration
was performed at the appropriate frequency using the appropriate number of standards. Review
the calibration summary results (e.g., Chapter 4.2.10) to ensure that acceptable initial calibra-
tions, CCVs, and ICVs were performed.
7-3.1. Goodness of Fit.

a. Check the linearity or goodness-of-fit for the calibration curve for each target ana-
lyte and surrogate by performing a qualitative visual examination of each calibration plot. Any
unusual problem observed during the visual examination of the calibration plots must be noted
(e.g., shallow slopes, high curvature at the upper of the calibration curve indicative of detector
saturation, large negative or positive intercepts, and large scatter).
b. When regression analysis is performed, it is especially critical to evaluate the fit for
the calculated curve near the origin. In particular, a high correlation coefficient can be obtained
(e.g., r > 0.995) when the calibration points significantly deviate from the curve.
c. Determine whether each of the reported regression coefficients or %RSDs are accept-
able. Using the calibration summary forms, recalculate the regression coefficient or %RSD (de-
pending on how the initial calibrations were performed) for at least one target analyte, and com-
pare the recalculated value to the reported value. If calibration was performed using the internal
standard technique, recalculate the relative response factors, mean response factor and %RSD
for one target analyte using the appropriate internal standard. The reported and calculated values
must agree to within two significant figures.
7-3.2. Representativeness of Initial Calibration Curve.
a. Verify that the initial calibration is representative of the analysis. In particular,
review the calibration summary forms to determine if any calibration points were erroneously
censored; that is, whether any calibration standards were inappropriately omitted from the
goodness-of-fit calculations (e.g., for the calculation of the %RSD or correlation coefficient).
Since instrumental response will be linear over a sufficiently narrow concentration range, the
standards at the extreme upper and lower ends of the calibration range (i.e., the highest and low-
est calibration standards) may be dropped to obtain a better linear fit. However, it is usually
inappropriate to drop calibration standards in the middle of the concentration range!
b. When any calibration standard between the highest and lowest initial calibration
standards is omitted, verify that a legitimate reason is documented (e.g., operator blunder, in-
strument malfunction, and the inadvertent use of an expired calibration standard). Censoring
calibration results (with the exception of the points corresponding to the extreme ends of the
concentration range) for no other reason than to obtain a better curve fit is an inappropriate
laboratory practice. An errant calibration result may actually be a valid result that reflects
higher-than-expected random analytical error. Under these circumstances, the omission of the
calibration result will under estimate the actual uncertainty associated with the environmental
samples.
7-4
EM 200-1-10
30 Jun 05
7-4. Qualification.
7-4.1. Frequency and Number of Standards.
If the instrument was not calibrated at the appropriate frequency with the appropriate number of
standards, qualify the associated sample results using professional judgement. For example, if
only four points were used to perform the initial calibration for the mercury analyses and a high
correlation coefficient was obtained, the associated mercury results would not typically be re-
jected. The width of the calibration range and the goodness of fit must be taken into account.
As a rule of thumb, at least three calibration standards are required when the quantitation range
is two orders of magnitude in width; additional standards are frequently required for larger cali-
bration ranges.
7-4.2. Representativeness.
When mid-level standards are inappropriately dropped, if possible, recalculate the calibration
curve and the goodness of fit (e.g., r2 or %RSD). If the goodness of fit is unacceptable, then
qualify the associated sample results as discussed in Chapter 7.4.3. If it is not possible or practi-
cal to recalcuate the curve and the goodness of fit, then, at a minimum, qualify all associated
detections with the J flag and all nondetections with the UN flag. The X flag may be appropriate
if the data are being used to support critical decisions.
7-4.3. Goodness of Fit.
a. Results must be qualified on the basis of quantitative acceptance limits for the calibra-
tion fits (e.g., the regression coefficients) and the visual examination of the initial calibration
plots. In particular, it may be necessary to qualify results when an acceptable coefficient of de-
termination or %RSD is obtained for the calibration. For example, if a nonlinear curve possesses
an acceptable coefficient of determination, it would be appropriate to qualify high concentration
samples if severe curve flattening from detector saturation were observed. Detections near the
intercept may be unreliable when a large nonzero y-intercept is obtained.
b. A conservative approach is recommended when method-specified minimum response

factors are not met. It is recommended that detections and nondetections be qualified with the X
flag or the R flag. (Minimum relative response factors for GC/MS analyses are typically 0.05 or
greater.) However, use professional judgement, and, at a minimum, qualify nondetections with
the UN flag and detections with the J flag.
c. Table 7-1 lists (to three significant figures) the maximum linear correlation coefficient
that is not significantly greater than zero as a function of the number of calibration points (n) and
confidence level (CL). For example, when the initial calibration is performed using five calibra-
tion points (e.g., for organic analyses), a correlation coefficient of 0.93 or less does not demon-
strate a positive linear correlation exists between instrument response and concentration at the
99% confidence level. Therefore (using the 99% confidence level), if the correlation coefficient
is less than or equal to 0.93, then detections and nondetections are qualified with the X or R flag.
Similarly, when the initial calibration is performed using four calibration points (e.g., three cali-
7-5
EM 200-1-10
30 Jun 05
bration standards and a blank for the metal analyses), a positive linear correlation is not demon-
strated at the 99% confidence level when the correlation coefficient is less than or equal to 0.98.
Table 7-1
Maximum r (Linear Correlation Coefficient) Not Significantly Greater Than Zero
Versus Number of Calibration Points for Initial Calibration 1
n 90% CL 95% CL 97.5% CL 99% CL

3 0.951 0.987 0.996 0.999
4 0.800 0.900 0.950 0.980
5 0.687 0.805 0.878 0.934
6 0.608 0.729 0.811 0.882
7 0.550 0.669 0.754 0.832
Notes: 1. CL denotes the confidence level and n denotes the number of points used to generate the initial calibration
line. The above table was generated using a one-tail t-test for the null hypothesis that the correlation coefficient is
not greater than zero
d. In the absence of project-specific requirements to do otherwise, qualify the results as

discussed below. It is assumed that at least five points (i.e., calibration standards) are used to
perform the initial calibration. When the initial calibration is performed using less than five
points, the acceptance limits for the %RSD and the correlation coefficient should be more strin-
gent.
e. When the initial calibration is performed using five or more calibration points, the ac-
ceptance criteria for the initial calibration are not met, and the correlation coefficient is greater
than 0.93 or the %RSD is less than or equal to 30% (when mean response factors are used to
perform the initial calibration), then, at a minimum, qualify nondetections with the UN flag and
detections with the J flag (as illustrated in Table 7-2). If the correlation coefficient is less than or
equal to 0.93 or the %RSD is greater than 30%, then qualify the results (i.e., the detections and
nondetections associated with the noncompliant target analyte) with the R flag. However, pro-
fessional judgement should be applied. For example, if the %RSD is grossly unacceptable, but a
review of the response factors for the initial calibration indicates that the %RSD is unacceptable
because of high variability at the high-end of the calibration line, it may be more appropriate to
qualify nondetections with the UN flag rather than the R flag.
f. Qualification of results for unacceptable calibration fits may be avoided when it is

possible to recalculate the calibration curve from the information enclosed in the data package.
For example, if the coefficient of determination or %RSD is unacceptable for a linear calibration
(e.g., because of curvature at the extreme low or high end of the calibration range), then drop-
ping the high or low end calibration standard may give an acceptable calibration line (i.e., an ac-
ceptable fit). The new calibration line may then be used to recalculate all the reported sample
results. However, at least three calibration standards are required to calculate a calibration line
7-6
EM 200-1-10
30 Jun 05
using regression analyses. Furthermore, since this strategy narrows the quantitation range, re-
sults that fall outside of the new calibration range would need to be qualified (as described be-
low).
Note: The degree to which recalculations will be required for the initial calibrations to
avoid data qualification will be highly situation-dependent. For example, it will be
function of the completeness of the data package, the level of effort negotiated for the
data review, and the degree to which QC requirements and corrective actions were
specified in the contract for laboratory analytical services.
Table 7-2
Data Qualification for Goodness of Fit for Organic Analyses with Five-
Point Linear Calibrations 1
Flag
Goodness of Fit Remarks Sample (y)
MRL < MQL < y None
r > 0.99
Acceptable MRL < y < MQL J
%RSD 15% Calibration y < MRL U
0.93 < r < 0.99 y > MRL J
Marginal
15% < %RSD 30% Failure y < MRL UN
r 0.93 y > MRL R
%RSD > 30% Gross Failure y < MRL R

Notes: 1. %RSD, r, MRL, MQL, and y denote the percent relative standard deviation of the response factors,
the linear correlation coefficient, method quantitation limit, method reporting limit, and the concentration of
analyte in the associated field sample, respectively. (The MRL is assumed to be greater than the MDL but
less than the MQL.)
7-4.4. Calibration Range.
a. All detections that fall outside of the calibration range of the instrument must be quali-
fied. For calibrations performed with multiple standards, the lowest and highest calibration
standards determine the lower and upper limits of the calibration range, respectively.
b. If the initial calibration is performed using a single (high) calibration standard for ICP
analyses, then the lower limit of the calibration range must be established by a low-level CCV
standard that is analyzed immediately after the initial calibration. If the initial calibration was
not verified via the analysis of a low-level CCV standard or if a low-level CCV was analyzed but
possesses a recovery that does not fall within 85% to 115%, then multiply the MDL by a factor
of ten (10) to establish the lower calibration limit (and the MQL). However, if the low-level
CCV is at least ten times greater than the MDL and the recovery does not fall within 85% to
7-7
EM 200-1-10
30 Jun 05
115%, then, at a minimum, qualify all detections between the low-level and mid-level CCVs as
estimated and qualify any associated nondetections with the UN flag.
c. As stated in Chapter 6, detections that are less than the lower limit of the calibration
range (e.g., the low calibration standard) must be qualified with the J flag. In addition, all detec-
tions that marginally exceed the high calibration standard must be qualified with the J flag; de-
tections that grossly exceed the upper calibration range must be qualified with the X flag (or the
R flag). The criteria for marginal versus gross failures must be determined using professional
judgement. The determinative technique as well as the range, fit, and shape of the calibration
curve must be taken into consideration. In general, if a sample result exceeds the upper calibra-
tion standard within the uncertainty tolerance for the CCV, it is recommended that the result be
qualified as a marginal failure. For example, if the high calibration standard is 100 ppb and the
CCV must be within 15% of its expected value, then detections greater than 100 ppb but less
than 115 ppb should be qualified with a J flag.
7-8
EM 200-1-10
30 Jun 05
CHAPTER 8
Initial Calibration Verification (ICV)
8-1. Introduction.
The initial calibration verification (ICV) is evaluated to assess the accuracy of the initial cali-
bration standards1.
8-2. Criteria.
The ICV must be performed after the initial calibration via the analysis of a mid-level standard.
The working calibration standards and the ICV standard must be from independent sources (e.g.,
from two different manufacturers). The recovery of the ICV should be within 90% to 110% for
inorganic analyses and within 80% to 120% for organic analyses.
8-3. Evaluation.
a. Review the standard preparation logs to verify that the ICV and initial calibration
standards were prepared from independent NIST-traceable standards. Review the instrument
printouts and run log sheets to verify that the ICV was analyzed after the initial calibration
within its expiration date. Using the standard preparation log sheets and the ICV summary form,
recalculate an ICV recovery and compare the calculated value with the reported value. If an ICV
standard was not prepared, review the standard preparation log to determine whether or not any
CCVs or LCSs were prepared from an independent-source standard
b. It should be noted that an ICV failure does not definitively demonstrate a source prob-
lem for the initial calibration standards. For example, failures may occur because of problems
with the initial calibration curve (e.g., a poor fit) and analytical blunder. Prior to qualifying the
data, it may be desirable to investigate the source of the failure (e.g., by requesting additional
information from the laboratory).
1
One could argue that an acceptable ICV does not definitively demonstrate the accuracy of the standards used for
the initial calibration. For example, the spiking concentrations for both the initial calibration and ICV standards
could be biased low (relative to the actual analyte concentrations in the standards). However, since both standards
should be traceable to a reliable source (e.g., NIST), an acceptable ICV supports the conclusion that the standards
are accurate. It is more likely than not that two different traceable standards that are in agreement are accurate.
8-1
EM 200-1-10
30 Jun 05
8-4. Qualification.
8-4.1. Frequency.
a. When an ICV is not performed (i.e., when initial calibration standards are not
verified with an independent-source standard) at a minimum, qualify all detections with the J
flag and all nondetections with the UN flag. Alternatively, the data review report must state that
all the results are potentially estimated. Rejection of the data may be appropriate when the data
is being used to support critical decisions.
b. The CCVs or the LCS may have been prepared from an independent-source standard.
If CCVs or LCSs are prepared from independent-source stock standards and the recoveries are
acceptable, then the data must not be qualified. If an independent-source standard for the ICV is
not commercially available, a standard from the same source material but a different preparation
lot (e.g., different manufacturers lot number) may be used for the ICV standard.
8-4.2. Percent Recovery.
8-4.2.1. Inorganics.
If the ICV recovery is unacceptable but falls within 80% to 120%, qualify detections with the J
flag and nondetections with the UN flag. If the ICV recovery does not fall within 80% to 120%,
then qualify the results with the X flag.
8-4.2.2. Organics.
If the ICV recovery is unacceptable but falls within 70% to 130%, qualify detections with the J
flag and nondetections with the UN flag. If the ICV recovery does not fall within 70% to 130%,
then qualify the results with the X flag.
8-4.3. Qualification for Bias.
When the ICV recovery is unacceptable or an independent-source standard is not used to verify
the initial calibration standard (e.g., an ICV is not performed), the direction of bias is unknown
for the entire analytical process. The recoveries of other QC samples (e.g., laboratory control
samples and matrix spikes) must not be used to make inferences about the direction of bias (e.g.,
unless the uncertainty is much greater than that arising from the ICV noncompliance.)
8-2
EM 200-1-10
30 Jun 05
Table 8-1
Data Qualification for ICV Results 1
ICV %R Method Remarks Sample (y) Flag

MRL < MQL < y None
90% %R 110% Inorganics
Acceptable MRL < y < MQL J
80% %R 120% Organics %R y < MRL U
110% %R 120%, 80% y > MRL J
%R 90% Inorganics
120% %R 130%, 70% Marginal

%R 80% Organics Failure y < MRL UN
%R > 120%, y > MRL X
%R < 80% Inorganics
%R > 130%,
%R < 70% Organics Gross Failure y < MRL X
Notes: 1. %R, MRL, MQL, and y denote the percent recovery of the target analyte in the ICV, method reporting
limit, method quantitation limit, and concentration of the target analyte in an associated field sample, respectively.
(The MRL is assumed to be greater than the MDL but less than the MQL.)
8-3
EM 200-1-10
30 Jun 05
CHAPTER 9
Continuing Calibration Verification (CCV)
9-1. Introduction.
Continuing calibration verifications (CCVs) are evaluated to determine whether the instrument
was within acceptable calibration throughout period in which samples were analyzed (i.e., to
verify that the initial calibration was applicable during the sample analyses). In general, failure
of the CCV indicates that the initial calibration is no longer valid and should trigger recalibration
and the reanalysis of the associated samples in the analytical sequence.
9-2. Criteria.
9-2.1. Traceability and Reporting Requirements.
The initial calibration and the sample analyses associated with each CCV must be clearly indi-
cated in the run log. The run log must also list the date each CCV standard was analyzed; the
time of analysis must also be specified for chromatographic methods. In addition, the source,
reference concentration (level spiked), measured concentration, and percent recovery must be
reported for each target analyte and surrogate (when surrogates are analyzed). However, for
chromatographic methods where the initial calibrations are performed using mean response fac-
tors, the percent differences for the CCV response factors may be reported instead of percent
recoveries (e.g., when the instruments software cannot readily report CCV recoveries).
CCVs must be analyzed in the same fashion as other QC samples (e.g., LCSs) and environmental
samples (i.e., must be analyzed in a manner that is representative of all other sample in the ana-
lytical sequence).
9-2.3. Frequency.
a. All environmental samples in an analytical sequence must be bracketed by (i) an

initial calibration and a CCV or (ii) by two CCVs. Therefore, a CCV must be analyzed at the
end of every analytical sequence.
b. If replicate CCVs are analyzed in succession before or after a set of samples, the
CCVs analyzed immediately before and after the samples constitute the bracketing pair of CCVs.
For example, Sample-01" and Sample-02" are qualified based upon the performance of CCV-
02 and CCV-03" for the analytical sequence:
CCV-01, CCV-02, Sample-01, Sample-02, CCV-03, CCV-04 ...
9-1
EM 200-1-10
30 Jun 05
c. However, it should be noted that a single reinjection of the CCV is typically

performed when a CCV fails. Therefore, if CCV-03 were to fail, the bracketing CCVs would
consist of CCV-02 and CCV-04.
9-2.3.1. Chromatographic Methods.
For chromatographic methods, a low-level or mid-level CCV standard must be analyzed at the
following frequency: (i) At the beginning of the analytical shift/sequence (when an initial cali-
bration is not being performed); (ii) every 12 hours of analyses or every 10 to 20 samples,
whichever comes first; and (iii) at the end of the analytical sequence--this includes GC/MS meth-
ods.
Note: The term sample refers to field samples and batch QC samples such as method
blanks, laboratory control samples, matrix spikes, matrix spike duplicates, matrix du-
plicates.
Note: Laboratories do not typically analyze a CCV at the end of the run sequence for
GC/MS analyses. In order to minimize the occurrence of estimated data, this require-
ment must be explicitly specified when contracting for laboratory analytical services.
Alternatively, a rationale for not analyzing CCVs at the end of the run sequence for
GC/MS analyses must be presented in project documents such as the QAPP.
For inorganic methods, a low-level or mid-level CCV must be analyzed at the following fre-
quency: (i) Every 10 to 15 samples and (ii) at the end of the analytical sequence.
9-2.4. Acceptance Criteria.
The acceptance limits for the CCVs will be highly dependent upon the analytical technique (as
well as the end use of the data). Therefore, several assumptions were made to develop the data
evaluation strategies that are presented in this section of the document. It was assumed that the
acceptance limits for the CCV are more stringent than the acceptance limits for the LCS when
the method of analysis involves significant sample preparation and the standards are not fully
processed with the environmental samples. Similarly, it was assumed that the CCV and LCS
limits will be similar when the method of analysis does not involve significant sample prepara-
tion. Lastly, a gross CCV failure was typically assumed to occur when a CCV exceeds twice
its tolerance for uncertainty.
a. If the method involves significant sample preparation and the CCVs are mid-level
standard solutions that are essentially instrument QC samples that are directly analyzed (e.g.,
CCVs for metals by GF-AA or ICP), then the recovery should be within 90110%. If the
calibration is verified using a CCV set at the low-level calibration standard, then the recovery
should fall within 85115%.
9-2
EM 200-1-10
30 Jun 05
b. Wider acceptance ranges should be used when the CCV is processed in the same
manner as the environmental samples (i.e., when the CCV is also an LCS) or when the CCV
undergoes a significant preparatory process. For example, CCVs are typically LCSs for the Hg
CV-AA analyses. When the CCV is processed in the same manner as the environmental
samples, then the CCV should be evaluated using the LCS acceptance limits; an acceptance
range of 80120% should be used. An acceptance range of 8515% is recommended when the
CCV is not processed in an identical manner as the samples but nevertheless undergoes a
significant preparatory process (e.g., cyanide CCVs that are distilled but that are not extracted
with the environmental samples).
9-2.4.2. Organic Chromatographic Methods.
a. If the calibration is verified with a mid-level CCV, then the recovery should be within
85115% for analyses of extractable organics (e.g., pesticides and Aroclors). If calibration is
verified using a CCV set at the low-level calibration standard, then the recovery should be 80
120%. For purge-and-trap methods (where the environmental samples and CCVs are prepared
and analyzed in the same manner), CCVs should be within 20% of their expected values. Wider
acceptance ranges may be appropriate for other organic methods where the CCVs are processed
in the same manner as the environmental samples.
b. Depending on the analytical method and the level of detail required for the evaluation,
additional acceptance criteria may be applicable for chromatographic methods. In particular, for
methods that require minimum response factors, the method-specified minimum response factor
criteria must be met. Methods such as 8260B and 8270C specify acceptance limits for the re-
sponses and retention times of the internal standards in the CCVs. The evaluation of internal
standards is discussed in Chapter 16. CCVs are often evaluated to determine if analyte identifi-
cation criteria are being met. In particular, for chromatographic methods involving the use of
two-dimensional detectors (e.g., FIDs and PIDs), CCV retention times are typically assessed to
verify that they fall within established retention time windows.
9-3. Evaluation.
a. Review the instrument run logs to verify that the CCVs were analyzed at an
appropriate frequency. Review the standard preparation log and note whether or not the CCVs
and initial calibration standards were prepared from the same source.
b. Use a continuing calibration summary form (and any instrument printouts of quantita-
tion reports) to recalculate a CCV recovery. For chromatographic methods where the initial
calibration is performed using mean response factors and percent differences are calculated for
response factors, calculate the percent difference for at least one response factor. Compare the
calculated values with the reported values. The former must agree with the latter to within at
least two significant figures.
9-3
EM 200-1-10
30 Jun 05
c. For each CCV, review the CCV summary form to verify that the reported percent
recovery or percent difference for each target analyte is acceptable. For chromatographic meth-
ods for which minimum response factors are specified, note any response factor that is not com-
pliant with method requirements.
9-4. Qualification.
a. Qualify the associated sample results if the CCVs were not analyzed in a
representative manner. In particular, the number of replicate analyses and system clean-out
activities must not be applied to CCVs to a greater extent than to the environmental samples in
the analytical sequence.
b. If the run sequence log indicates that multiple CCBs (continuing calibration blanks)
are analyzed before the CCVs but not before any of the environmental samples, then the CCVs
may not be representative. If the replicate CCB analyses were being performed to address carry
over, then qualify the associated sample results as estimated or rejected depending upon the se-
verity of the blank contamination and the intended use of the data. Ideally, the laboratory should
be required to provide the entire raw data package and the CCB with the highest level of carry
over (typically the first CCB in the run sequence) should be used to qualify the associated sam-
ple results for blank contamination using the strategies in Chapter 10. However, when the CCB
results are not available, at a minimum, qualify all detections in the associated environmental
samples as estimated (with the J+ flag).
c. If multiple CCVs are being analyzed, the representativeness of the CCV results must
be critically evaluated. For example, assume that the following run sequence is observed for a
set of aqueous VOC analyses:
CCV-01, CCV-02, CCV-03, CCV-04, Sample-01, Sample-02, MB, CCV-05, CCV-06,

CCV-07 ...
d. Assume that CCV-04 and CCV-07 are acceptable (i.e., the CCV recoveries fall within
the acceptance range), but the remaining CCVs are unacceptable. Although two acceptable
CCVs bracket the samples, the run sequence suggests that the CCVs are not being analyzed in an
appropriate (i.e., representative) manner. When analytical problems exist (especially when a
method is only marginally out-of-control), if a sufficient number of QC samples (such as CCVs)
are analyzed, then one of the QC samples will eventually fall within the acceptance limits by
chance (i.e., because of random error)! Method performance appears to be acceptable but is ac-
tually substandard (most of the CCVs are not falling within the acceptance limits). Under these
circumstances, qualify the associated sample results (e.g., Sample-01 and Sample-02) using the
most noncompliant CCV recovery. If this information is not available (e.g., the recoveries for
only CCV-04 and CCV-07 are reported), then qualify all the associated sample results for mar-
ginal CCV failure (refer to Chapter 9-4.4). If the data are being used to support critical deci-
sions, it may be appropriate to qualify the sample results as tentatively unusable (using the X
flag).
9-4
EM 200-1-10
30 Jun 05
c. CCVs are occasionally used to update the instruments calibration data (e.g.,
resloping for GF-AA analyses). This is not the objective of a CCV. A CCV is performed to
verify (to within some tolerance for uncertainty) that the initial calibration remains valid and is
not performed to alter the initial calibration curve. Updating the calibration using the CCV
primarily amounts to replacing the original multiple-point calibration with a single-point calibra-
tion. When this occurs, recalculate the associated results using the original calibration curve or,
at a minimum, qualify the results as estimated. However, professional judgement should be
used. For example, when there is significant instrumental drift and a calibration line is updated
using the CCV, results calculated from the CCV (particularly mid-range detections) may be
more accurate than those calculated from the multiple-point calibration!
9-4.2. Frequency.
a. If a CCV is missing at the end of the analytical sequence, then, at a minimum, qualify
all detections with the J flag and all nondetections with the UN flag unless it can be otherwise
demonstrated that the instrument remained in calibration for the entire analytical sequence. For
example, the laboratory may have analyzed extremely dirty environmental samples near the
end of the run sequence and cleaned the instrument to eliminate carry over problems only for
the next 12-hour CCV. Qualification of the associated sample results with the X flag may be
more appropriate for some data uses (e.g., when the data is being used to support critical deci-
sions).
b. If all samples are bracketed by two acceptable CCVs but the CCVs are not analyzed at
the appropriate frequency (e.g., after every 10 to 20 samples), use professional judgement to de-
termine whether data qualification is necessary. For significant nonconformances, qualify de-
tections with a J flag and nondetections with the UN flag.
9-4.3. Tolerance for Uncertainty.
a. In general, if a CCV in an analytical sequence is not acceptable, then qualification is

required for all samples following the last acceptable CCV and all samples preceding the next
acceptable CCV. For example, consider the following run sequence:
CCV-01, Sample-01, Sample-02, CCV-02, Sample-03, Sample-04, CCV-03, Sample-

05, Sample-06, CCV-04 . . .
b. Sample-01" to Sample-04" would be qualified if CCV-02 were unacceptable.

Qualification protocols for CCV failures are very similar to those for LCS failures. Marginal
CCV failures are distinguished from gross failures as discussed below.
9-5
EM 200-1-10
30 Jun 05
9-4.3.1. Inorganic Methods, CCVs Not Processed with Samples.
If the CCV does not undergo a significant preparatory process relative to the environmental
samples, then evaluate the CCV results as follows: If the CCV recovery is unacceptable but falls
within 80% to 120%, then qualify the data (i.e., the associated sample results) for marginal fail-
ure. If the CCV recovery is unacceptable and does not fall within 80% to 120%, then qualify the
data for gross failure.
9-4.3.2. Inorganic Methods, CCVs Processed with Samples.
If the CCV is processed in the same manner as the environmental samples, then the CCV is es-
sentially an LCS and must be evaluated using the LCS limits. The results should be qualified for
marginal failure if the CCV is unacceptable but falls within 60% - 140% of the expected value.
If the CCV undergoes a significant sample preparatory process but is not processed in an identi-
cal manner as the environmental samples, then it is recommended that results be qualified for
marginal failure if the CCV is unacceptable but falls within 70% - 130% (e.g., cyanide CCVs
that are distilled but not extracted with the environmental samples).
9-4.3.3. Organic Methods, CCVs Not Processed with Samples.
The following guidance applies to methods that require significant sample preparation (e.g., sol-
vent extractions or cleanup procedures) and the CCV is not processed with the environmental
samples. If the CCV is unacceptable but the percent recovery falls within 70% to 130% or the
percent difference for the response factor is not greater than 30%, then qualify the data for mar-
ginal failure. If the CCV is unacceptable and the percent recovery does not fall within 70% to
130% or the difference for the response factor is greater than 30%, qualify the associated sample
results for gross failure.
9.4.3.4. Organic Methods, CCVs Processed with Samples
If the method does not require significant sample preparation or the CCV is processed with the
samples (e.g., aqueous purge-and-trap analyses), the CCV is unacceptable but the percent recov-
ery falls within 40% to 160% or the percent difference for the response factor is not greater than
60%, then qualify the data for marginal failure. If the CCV is unacceptable and the percent re-
covery does not fall within 40% to 160% or the difference for the response factor is greater than
60%, then qualify the associated sample results for gross failure.
9.4.4. General Qualification Strategies.
a. Environmental sample results are qualified for CCV failure, based upon the (i) direc-
tion of bias, (ii) the magnitude of the failure, and (iii) the concentration of the target analyte
relative to the AL. The direction of bias for a CCV failure is well defined when all other associ-
ated QC samples (e.g., ICVs and LCSs) are in control or exhibit bias in the same direction, i.e., if
the CCV recovery is unacceptably high but the LCS recovery is unacceptably low, then the di-
rection of bias is not well defined. Similarly, if the ICV is unacceptable or if a second source
standard was used to prepare the CCV and the CCV is unacceptable, then the direction of bias
cannot be inferred from the CCV recovery. Qualification strategies for CCV failures follow.
9-6
EM 200-1-10
30 Jun 05
(1) If the CCV is marginally unacceptable and the direction of bias is well defined, then
the data is qualified as follows: For low bias, qualify detections with the J- flag and
nondetections with the UN flag. For high bias, qualify detections with the J+ flag and
nondetections with the U flag.
(2) If the CCV is marginally unacceptable and the direction of bias is not well defined,
then qualify detections with the J flag and nondetections with the UN flag.
(3) If the CCV is grossly unacceptable and the direction of bias is well defined, then
qualify the associated sample results as follows:
(a) For low bias, qualify all nondetections with the R flag. When an AL is not specified,
qualify detections with the J- flag. If an AL is specified, then qualify detections less than the AL
with the X flag and qualify detections greater than the AL with the J- flag.
(b) For high bias, qualify all nondetections with the U flag. Qualify detections with the
J+ flag. However, when an AL is specified, it may be appropriate to qualify detections greater
than the AL with the X flag. Alternatively, it may be desirable to obtain additional information
from the laboratory before completing the evaluation. For example, additional data could be re-
quested to determine if the high CCV recovery resulted from carry over or improper integra-
tions
(4) If the CCV is grossly unacceptable and the direction of bias is not well defined, then
qualify nondetections with the R flag. When an AL is not specified, qualify detections with the J
flag. If an AL is specified, qualify detections with the X flag. (However, if possible and practi-
cal, the magnitude of the uncertainty relative to the proximity of the detection to the AL should
be taken into account.)
b. The qualification strategies discussed above are illustrated in Table 9-1 (where it is as-
sumed that each CCV must be within 10% of its expected values). However, CCV failures must
be interpreted in the context of other instrumental and batch QC results using professional
judgement. In particular, a result may still be acceptable when an associated CCV does not fall
within the CCV acceptance limits because the uncertainty tolerance for instrumental perform-
ance is typically more stringent than that for overall method performance. For example, if the
CCV recovery must be within 10% of its expected value and the LCS must be within 20% of its
expected value, but the CCV recovery is 85% and the LCS recovery is 80%, then overall accu-
racy of the associated sample results is still acceptable. In general, if the direction of bias is well
defined and the LCS is in control, sample qualification is not required when the CCV recovery is
marginally unacceptable. (However, under these circumstances contractual corrective action
may be appropriate.)
9-7
EM 200-1-10
30 Jun 05
Table 9-1
Data Qualification for CCV Results 1
%R for CCV Bias Remarks Sample (y) Flag

MRL < MQL < y None
MRL < y < MQL J
Acceptable
90% %R 110% %R y < MRL U
y > MRL J
110% %R 120% or Marginal
80% %R 90% Failure
Undefined Bias y < MRL UN
Marginal y > MRL J-
80% %R 90% Failure
Low Bias y < MRL UN
110% %R 120% Marginal y > MRL J+
High Bias Failure
y < MRL U
X if y < AL
y > MRL J- otherwise
%R < 80%
Low Bias Gross Failure y < MRL R
J+
y > MRL Possibly, X if y > AL
%R > 120%
High Bias Gross Failure y < MRL U
J if AL not specified;
%R > 120 or y > MRL X if AL specified
%R < 80%
Undefined Bias Gross Failure y < MRL R
Notes: 1. %R, MRL, MQL, AL, and y denote the percent recovery of the target analyte in the CCV, method report-
ing limit, method quantitation limit, AL, and concentration of the target analyte in an associated field sample,
respectively. It is assumed that the MRL is greater than the MDL, less than the MQL, and less than the AL.
9-8
EM 200-1-10
30 Jun 05
CHAPTER 10
Blanks
10-1. Introduction.
Blanks are assessed to determine the existence and magnitude of contamination problems and
measure of the representativeness of the analytical process. Blanks reflect the amount of con-
tamination introduced into the environmental samples during sample collection, transfer or
analysis. In particular, method blanks reflect laboratory contamination from both the determi-
native and preparatory method. Field blanks (e.g., trip blanks and equipment or rinsate blanks)
account for accumulative field and laboratory activities. In general, the samples associated with
each blank (e.g., method and field blanks) must not be corrected for blank contamination (e.g.,
unless QAPP or the method of analysis describes a valid procedure for correcting for blank con-
tamination).
Note: Although blank contamination imparts a high bias to analytical results, blanks are
not viewed to be a measure of bias or positive interference because a one-to-one cor-
respondence between blank contamination and bias does not exits (e.g., high LCS re-
coveries can be obtained when contamination is not detected in any blanks). Blank
contamination is indicative of an effect that is external to the native sample matrix and
relates the representativeness of the sample.
10-2. Criteria.
10-2.1. Frequency.
a. At least one method blank must be reported for each preparation batch of samples.
Note: Method blanks associated with a set of environmental samples must be analyzed
with the environmental samples using the same instrument in the sample analytical run
sequence. For example, if a batch of 20 samples is prepared with a method blank, some
of the environmental samples are analyzed with the method blank on day one, and
the remaining environmental samples are analyzed on day two, then the same method
blank analyzed on day one should be analyzed on the second day of analysis. At a
minimum, an instrument blank must be analyzed with the remaining environmental
samples on day two.
b. Trip blanks must be reported for each cooler containing VOC samples. Additional
field blanks may be required for certain projects. The frequency of collection and types of field
blanks must be evaluated against project-specific requirements.
10-2.2. Acceptance Limits.
The concentration of each target analyte in each blank must be less than the greater of the fol-
lowing: (i) the RDL for the target analyte; (ii) the MRL when the MRL is not greater than 5% of
10-1
EM 200-1-10
30 Jun 05
the AL, (iii) 5 to 10% (depending on project DQOs) of the analyte concentration detected in each
associated field samples; and (iv) 5 to 10% (depending on project DQOs) of the AL. Environ-
mental sample detections greater than the MRL but less than 10 times the corresponding blank
detections must be qualified. In instances where more than one blank is associated with a given
sample (e.g., a rinsate blank and method blank), evaluate blank contamination using the associ-
ated blank containing the highest contaminant concentration
Note: Laboratories commonly set the method blank acceptance criteria at the method
reporting limit (MRL), which in turn is set equal to the method quantitation limit
(MQL). This is not appropriate when action level is near the MRL/MQL! When blank
acceptance criteria are established based upon the MRLs, blank contamination between
the RDLs and MRLs must be reported when the MRL is greater than 5% of the AL.
10-3. Evaluation.
a. Review the Case Narrative and note any problems with method blank contamination.
Review the summary forms for method blanks and any field blanks (e.g., trip blanks and rinsate
blanks). Significant contamination in a blank may be an isolated occurrence. However, if the
reviewer cannot reasonably demonstrate that a contamination problem is an isolated occurrence,
a conservative approach must be used. Qualify the environmental sample results using the high-
est analyte concentration detected in the associated blanks (e.g., the method, field, and instru-
ment blanks).
b. Although data qualification strategies for blank contamination are presented in

Chapter 10.5, professional judgement is also required. Factors such as the magnitude and
frequency of the blank contamination, the nature of the site contamination, the nature of the
analysis, and historic data regarding the presence of blank contaminants should also be taken
into account. For example, assume that methylene chloride has not been detected during prior
sampling efforts (e.g., long-term groundwater monitoring) and methylene chloride has been
historically detected in a sporadic manner in associated blanks at low-levels. Furthermore,
assume that two batches of groundwater samples are reported for the most current sampling
event, Batch 1" and Batch 2." Methylene chloride is detected at low levels in the
environmental samples in Batch 1" and Batch 2, but methylene chloride is detected only in
the method blank for Batch 1. It would be reasonable to qualify the low-level methylene
chloride detections for the samples of Batch 2" on the basis of the method blank associated with
Batch 1," even though all the blanks associated with Batch 2" are clean.
10-4. Contractual Considerations.
a. Since laboratories are normally required to reprocess (e.g., reextract and reanalyze) a
batch of samples when the method blank is unacceptable, contractual corrective action for un-
satisfactorily performance may be warranted when high levels of contamination are systemati-
cally observed in the method blanks or when a method blank is not processed. Similarly, con-
tractual corrective action may be appropriate for unacceptable field blanks (e.g., rinsate and field
blanks).
10-2
EM 200-1-10
30 Jun 05
b. When high blank contamination is observed, the reviewer should consult with the Pro-
ject Manager to determine whether the data package must be reviewed or rejected. For example,
the laboratory may be required to reanalyze the environmental samples. Alternatively, it may be
possible to adopt higher reporting limits (e.g., when the higher reporting limits are still much
lower than the projects decision limits).
Note: Meeting the method blank acceptance criteria on a routine basis may not be
practical for common laboratory contaminants (e.g., methylene chloride, phthalates,
and acetone); sporadic detections of contamination may occur and are difficult to con-
trol. Exercise professional judgment when evaluating contractual compliance for
common laboratory contaminants.
10-5. Qualification for Blank Contamination.
a. When a target analyte is detected above the RDL in any blank, qualification for the
associated environmental samples for blank contamination is not required when any of the fol-
lowing occur:
(1) The target analyte is not detected in the environmental samples.
(2) The target analyte is detected in the blank at a concentration less than 5% to 10% of
the corresponding environmental sample concentration.
(3) The target analyte is detected in the blank at a concentration greater than the RDL
and less than the MRL, where the MRL is less than 5% of the AL.
b. In general, qualification is required when a target analyte is detected in a blank at a

concentration greater than 5 or 10% of the corresponding environmental sample concentration
(e.g., even when the analyte is detected at less than 5% of the AL). Qualification for blank con-
tamination is illustrated in Table 10-Samples are qualified for blank contamination using the
following strategies:
(1) J+ flag. If the analyte concentration for an environmental sample is greater than five
but less than ten to twenty times higher than the analyte concentration in the corresponding
blank, qualify the reported sample result with a J+ flag. Under these circumstances, the J+ flag
indicates that the analyte is present in the sample but the reported concentration of the analyte
believed to be biased high because of blank contamination. When the analyte concentration for
an environmental sample is less than five times the analyte concentration in an associated blank,
data qualification will be highly dependent upon project-specific DQOs. In particular, qualifica-
tion will be dependent upon whether or not action levels are available. Sample results are quali-
fied with the U, UN, X, or N flag as discussed below.
(2) UN flag. If the analyte concentration for the environmental sample is less than five
times the analyte concentration in the corresponding blank, then qualify the sample result with
the UN flag if (i) an AL is not available or (ii) the sample result is less than the AL. The UN flag
indicates that the analyte was not reliably detected because of blank contamination and the re-
10-3
EM 200-1-10
30 Jun 05
ported result is viewed as a tentative nondetection at the reported concentration. Alternatively,

multiply the blank by a factor of five and report (in place of the sample result) the resulting
product with a U flag when (i) the product is significantly less than the AL (e.g., 5% or 10% of
the AL) or (ii) an AL is not available.
(3) X flag. If the analyte concentration for the environmental sample is less than five
times the analyte concentration in the corresponding blank but is greater than the AL, then
qualify the sample result with the X flag. Under these circumstances, the X flag indicates that
the analyte was not reliably detected (above the AL) because of blank contamination and should
be rejected. In effect, blank contamination has increased the reporting limit for the analyte to a
concentration that is greater than the AL. A nondetection reported at the elevated limit does not
demonstrate the target analyte is present in the environmental sample above or below the AL.
The sample result must not be qualified with the UN (or U flag) unless a defensible technical
rationale for the use of the flag UN is presented.
(4) N flag. If the analyte concentration for the environmental sample is less than five
times the analyte concentration in the corresponding blank and the analyte concentration is
greater than the AL, then qualify the sample result with an N flag only when it can be demon-
strated that the UN flag or X flag is not appropriate. For example, the N flag may be appropriate
when it is desirable to establish an upper limit for site-related contamination. When used in this
manner, the N flag indicates that a target analyte result is being reported as a detection but the
detection may not be reliable because of contamination problems.
10-4
EM 200-1-10
30 Jun 05
Table 10-1
Data Qualification for Blank Contamination 1
Remarks
Blank Reported Qualified AL = 100 ppb Blank acceptance
(BLK) Result (y) Result MRL = 1 ppb criteria in Chapter
(ppb) (ppb) (ppb) MQL = 5 ppb 10.2.2 met?
No significant Yes
1U 6 6 contamination detected. BLK < RL < 5% of AL
Contamination detected, No 3
23 <1 1U but no action required. BLK > 5% of AL
4 UN or y < 5 BLK Yes
2J 4 10 U and y < AL 2 BLK < 5% of AL
5 BLK < y < 20 BLK Yes
2J 11 11 J+ BLK < 5% of AL
y > 20 BLK Yes
2J 60 60 BLK < 5% of AL
No 4
150 X y < 5 BLK BLK > 5% of y
80 150 and y > AL
Notes: 1. The concentration of analyte detected in the field sample and blank are denoted by y and BLK, respec-
tively. For the purposes of illustration, it is assumed that MRL = 1 ppb, MQL = 5 ppb, and AL = 100 ppb, where
RDL MRL. Note that the MRL is less than 5% of the AL. 2. The same flags would be applied to the sample
result if an AL were not available. 3. Although acceptance criteria for blank contamination in Chapter 10.2.2 were
not met, the result is still usable. The laboratory would not typically be required to reprocess the sample for method
blank contamination but should be expected to investigate the source of the contamination. 4. The laboratory would
typically be required to reprocess the sample for method blank contamination
10-5
EM 200-1-10
30 Jun 05
CHAPTER 11
Laboratory Control Samples (LCSs)
11-1. Introduction.
Laboratory control samples are evaluated to assess overall method performance and are the
primary indicators of laboratory performance. In general, laboratory control samples are similar
in composition as the environmental samples, contain known concentrations of all the analytes
of interest, and undergo the same preparatory and determinative procedures as the environmental
samples. LCS recoveries are used to measure accuracy. The relative percent difference (RPD)
for duplicate LCS recoveries is normally used as a measure of precision. When both a laboratory
control sample (LCS) and laboratory control sample duplicate (LCSD) are processed for a batch
of samples, there is no significant physical distinction between the LCS and LCSD. Both the
LCS and LCSD must satisfy the same recovery acceptance criteria. Therefore, for simplicity,
the term LCS will refer to one or more laboratory control samples (e.g., the term LCS accep-
tance criteria will refer to the acceptance criteria for both the LCS and LCSD).
11-2. Criteria.
11-2.1. Frequency.
At least one LCS must be reported with each batch of samples. A laboratory control sample and
a laboratory control sample duplicate (LCSD) may be analyzed to provide information on the
precision of the analytical method. The generation of control chart limits for precision via the
analysis of LCS/LCSD pairs is an effective means to measure method precision. Multiple LCSs
may be required to evaluate method precision and accuracy at different spiking concentrations.
a. Project documents such as the QAPP should specify the acceptance limits for LCS
recoveries. To the extent possible, LCS acceptance limits should be established based upon
project DQOs rather than upon contractual specifications, the limitations of the laboratory, or the
limitations of the analytical method. Laboratory statistical control limits should be evaluated
during the planning stages of the DQO process to assure that project-required acceptance limits
will be met.
b. Laboratory statistical control limits must not be the sole basis upon which project-
required acceptance limits are established. Statistical control limits generated by the laboratory
may be representative of routine method performance but may be too wide to satisfy project-
specific DQOs. Furthermore, statistical control limits for laboratory control samples tend to ad-
versely impact laboratory-to-laboratory comparisons (e.g., when USACE QA split sample
analyses are being performed, an LCS recovery that falls within the wide acceptance range of
one laboratory will not necessarily fall within the tighter acceptance range of the referee labora-
tory or vice versa).
11-1
EM 200-1-10
30 Jun 05
c. Acceptance limits for bias and precision are presented in various analytical methods
(e.g., SW-846 and CLP methods) but many of these limits may be inappropriately wide. Accep-
tance limits for accuracy and precision are presented in the USACE Shell. Although these limits
were established to ensure a moderate to high level of data quality, they are ultimately contrac-
tual in nature (e.g., permit poor performance for select target analytes because of inherent limi-
tations of the analytical methodology). It may not be practical or possible (even after method
modification and development) for a method to routinely meet the acceptance limits for every
target analyte. Under these circumstances, the reviewer must distinguish contractual compliance
and laboratory performance issues from data usability issues.
d. Inappropriately wide LCS acceptance ranges may be specified for a method in project-
documents such QAPPs, SAPs, and Work Plans. These acceptance ranges are often based upon
contractual, method-specified, or laboratory control chart limits. For example, erroneously wide
LCS acceptance ranges may be specified when ALs are equal to or near the MQLs. The specifi-
cation of an acceptance limit in a project document per se does not imply that limit is scientifi-
cally sound with respect to project objectives. When, in the reviewers professional judgment,
project-specified LCS acceptance limits are not consistent with project DQOs, evaluate the data
package with respect to scientifically defensible limits.
e. In the absence of reasonable LCS recovery limits, the following limits are recom-
mended: The recovery for each target analyte should fall within 80 to 120% for inorganic analy-
ses and within 60 to 140% for organic analyses. For purge-and-trap GC and GC/MS analyses,
recoveries should fall within 80% to 120% when the CCV is being used as the LCS. If the LCS
is an independent source standard, the LCS should fall with 70 to 130% for purge-and-trap
analyses.
f. In the absence of project-specific limits for precision, it is recommended that the

acceptance limit for the RPD be equal to one half of the width of the corresponding LCS
recovery acceptance range or to the laboratorys RPD acceptance limit, whichever is less. For
example, the laboratory may have established statistical RPD acceptance limits by processing an
LCS/LCSD pair for each batch or from interbatch LCSs (i.e., LCSs from consecutive batches).
11-3. Evaluation.
Evaluate the LCS results using the following strategies:
a. Using the standard preparation logs verify that all target analytes were spiked into the
LCS and note whether or not an independent-source standard was used to prepare the LCS.
Note: A number of published analytical methods do not require the LCS to contain all
the target analytes. Unless a scientifically defensible rationale for not spiking all the
target analytes is presented in the analytical method or in project documents such as the
QAPP, assume that all single-component target analytes must be spiked into the LCS.
However, when several multi component target analytes are being simultaneously ana-
lyzed (e.g., the set of Aroclors in Method 8082), it may not be possible (or desirable) to
spike all the analytes into a single LCS. Depending on the nature of the analysis and
11-2
EM 200-1-10
30 Jun 05
the data quality objectives for the project, a set of laboratory control samples (e.g., one
LCS for each multi component target analyte) may be required or only a single LCS
containing representative components may be appropriate (e.g., an LCS containing
Aroclors 1016 and 1260 is typically assumed to be representative of the other Aroclors
analyzed by Method 8082).
b. Using the sample preparation log and the instrument run log verify that the LCS was
processed with the samples through the entire analytical method.
c. Using the LCS summary form, calculate the LCS recovery for at least one target ana-
lyte and compare the calculated value to the reported value. Similarly, recalculate the RPD for
an LCS/LCSD pair for one target analyte and compare the calculated value to the reported value.
The calculated LCS recoveries and RPDs must agree with the reported values to within two sig-
nificant figures.
d. For each target analyte, compare the LCS recoveries and RPDs reported on the labora-
torys summary forms to the corresponding LCS acceptance limits for bias and precision. In the
absence of appropriate acceptance limits, establish a set of limits to properly evaluate the LCS
results. A batch of samples is acceptable only for those target analytes that satisfy the LCS crite-
ria for bias and precision. All failures must be noted. Data qualification is required when the
LCS acceptance criteria are not met.
e. Review the Case Narrative and note any problems discussed for the LCS. When an
LCS recovery is unacceptable, examine the Case Narrative and note why the batch was not re-
processed (e.g., reextracted and reanalyzed) for the failed analyte. However, it should be noted
that even when method implementation is optimal, a small percentage of sporadic failures should
be expected for the LCS (especially when a large number of target analytes are being simultane-
ously analyzed).
a. Contractual considerations may impact the data review. Since laboratories are nor-
mally required to reprocess (e.g., reextract and reanalyze) a batch of samples when the LCS is
unacceptable, contractual corrective action for unsatisfactorily performance is typically required
for gross systematic LCS failures. When gross systematic failures occur, the reviewer should
consult with the Project Manager to determine whether or not to proceed with the review or to
reject the data package as a whole (e.g., the laboratory may be required to reanalyze the envi-
ronmental samples). However, the reviewer should exercise professional judgment when de-
termining whether contractual compliance will impact the data review. In particular, for meth-
ods containing large lists of target analytes (e.g., Method 8270C) or poor performers (e.g., the
ketones of Method 8260B or other analytes which cannot meet QC limits because of inherent
method limitations), it is highly probable that the recoveries of several target analytes will be un-
acceptable.
b. Sporadic marginal LCS failures should be expected and should not trigger a consulta-
tion with the Project Manager or the rejection of a batch of samples. For example, a marginal
11-3
EM 200-1-10
30 Jun 05
sporadic failure may be said to exist if an LCS recovery falls between the three- and four-sigma
control limits for no apparent reason for a particular batch of samples but the laboratory control
samples for prior and subsequent batches are acceptably recovered. The table below lists the
maximum number analytes expected to fall outside of the three-sigma control limits for an LCS
when the LCS contains a large set of target analytes..
c. For example, according to Table 11-1, if there are 20 target analytes, as many as two
analytes in the LCS may fall outside of the three-sigma acceptance limits because of random
error. Typically, these types of sporadic failures should not trigger reanalyses of the batch but
the associated environmental sample results should be qualified.
Table 11-1
Number of Target Analytes Versus Number of Expected LCS Failures
n1 f2
1015 1
1645 2
4685 3
86130 4
Notes: 1. n = Total number of target analytes being simultaneously analyzed. 2. f = Maximum number of analytes
expected to fall outside of the three-sigma control limits with 99% confidence if the probability of a random failure
is less than or equal to 1%.
Note: Review project documents (e.g., the Quality Assurance Plan) to ensure that the
noncompliant analyte is not a critical analyte (e.g., a human or ecological risk driver).
For example, if 60 VOCs are being analyzed by 8260B, but vinyl chloride is the pri-
mary contaminant of concern, then reanalyses for vinyl chloride should be expected
when the LCS recovery is not acceptable.
d. If precision is unacceptable for a particular analyte (e.g., the RPD is higher than the
acceptance limit), then the associated field sample detections above the MQL (or the MRL if it is
greater than the MQL) must be qualified as estimated data. To satisfy project-specific require-
ments, the laboratory may be required to reprocess a batch of samples when the LCS does not
satisfy precision acceptance criteria. Under these circumstance, verify that this was done. How-
ever, it should be noted that laboratories do not typically reprocess environmental samples for
unacceptable RPDs when the LCS recoveries are acceptable.
11-5. Qualification.
a. The qualification strategies presented in this section of the document will generally be
applicable.
11-4
EM 200-1-10
30 Jun 05
(1) When multiple laboratory control samples (e.g., an LCS and LCSD) are processed
for a single batch of samples, and one or more LCS recoveries are unacceptable for a particular
target, then the associated samples must be qualified on the basis of the most noncompliant
target analyte recovery. However, it should be noted that replicate laboratory control samples
may not be required or reported. For example, if the RPD for an LCS/LCSD pair is calculated
using interbatch laboratory control data (i.e., the LCSD is not extracted with the LCS but is the
control sample for a consecutive batch of samples), the LCSD recovery may not have been
reported..
(2) Data qualification must be a function of both the magnitude and direction of the QC
failure. Gross QC failures must be distinguished from marginal failures and the direction of bias
must be taken into account. When the LCS recovery is unacceptable, the direction of bias will
be said to be well defined if the direction of bias for other batch and instrument QC samples
(e.g., ICVs, surrogates, and replicate LCSs) is consistent with the noncompliant LCS recovery.
For example, if both an LCS and LCSD are extracted with a batch of samples and the LCS re-
covery is less than the lower control limit but the LCSD recovery is greater than the upper con-
trol limit, then the direction of bias is not well defined. Similarly, the direction of bias is not
well defined when the RPD for an LCS/LCSD pair is used to evaluate duplicate precision and
the RPD is unacceptable, but the LCSD recovery is not reported
b. Specific qualification protocols for laboratory control samples are presented below
and are illustrated in Table 11-2 (where it is assumed that all QC samples other than the LCSs
are in control).
(1) If the LCS recovery is marginally unacceptable and the direction of bias is not well
defined, then qualify detections of the target analyte with the J flag and nondetections with the
UN flag.
(2) If an LCS recovery is marginally unacceptable and the direction of bias is well de-
fined, then qualify the data as follows: For low bias, qualify detections with the J- flag and non-
detections with the UN flag. For high bias, qualify detections with the J+ flag and nondetections
with the U flag.
(3) If an LCS recovery is grossly unacceptable and the direction of bias is well defined,
then qualify the associated sample results as follows:
(a) For low bias, qualify all nondetections with the R flag. If an AL is not specified,
with the X flag and detections greater than the AL with the J- flag.
than the AL with the X flag (e.g., when a conservative estimate is not being sought).
(4) If the LCS recovery is grossly unacceptable and the direction of bias is not well de-
fined, then qualify nondetections with the R flag. If an AL is not specified, then, at a minimum,
11-5
EM 200-1-10
30 Jun 05
qualify detections with the J flag (the X flag may be more appropriate). If an AL is specified,
qualify detections less than the AL with the X flag. Depending on project DQOs, qualify detec-
tions greater than the AL with the J or X flag.
c. In addition to the qualification strategies discussed above, use the following protocols
when duplicate laboratory control samples are processed with each batch of samples:
(1) If the LCS/LCSD recoveries are acceptable, the RPD is marginally unacceptable, and
the direction of bias is not well defined, then qualify detections with the J flag and nondetections
with the UN flag.
(2) If the RPD is grossly unacceptable and the direction of bias is not well defined, then
qualify nondetections with the R flag. (The X flag may be appropriate if additional information
to determine the direction of bias will be obtained). Qualify detections with the J flag when an
AL is not specified. If an AL is specified, then qualify detections less than the AL with the X
flag and qualify detections greater than the AL with the J flag or the X flag.
d. In the absence of valid project-specific limits for bias and precision, a gross failure is
defined to occur when one of the following conditions is satisfied:
(1) For inorganic analyses, a gross failure occurs for a target analyte when the percent
recovery does not fall within 60 to 140%. For organic analyses involving significant sample
preparation (e.g., solvent extraction), a gross failure occurs when the LCS recovery does not fall
within 20 to 180%. However, for purge-and-trap analyses, a gross failure occurs when the LCS
recovery does not fall within 40 to 160%.
(2) A gross failure occurs when the RPD for the LCS/LCSD is greater than 40% for
inorganic analyses, 60% for purge-and-trap analyses, and 80% for extractable organic analyses.
11-6. Qualification Strategies Using Estimates of the Uncertainty.
a. This section of the document describes some optional data qualification strategies that
may be used when analytical uncertainty can be estimated from laboratory control samples.
These strategies will be applicable when matrix interference and sample heterogeneity are not
significant components of the analytical uncertainty or when it is desirable to establish a lower
bound for the total uncertainty. Laboratory uncertainty is estimated from the laboratorys in-
house statistical warning and control limits for LCS recoveries. If representative matrix spike
warning and control limits are available, it is recommended that these limits be used instead of
the LCS limits. The use of matrix spike warning and control will result in better estimates of the
uncertainty (e.g., since LCS limits do not account for the uncertainty associated with matrix ef-
fects). However, it should be noted that representative matrix spike recovery limits are not typi-
cally available from environmental production laboratories and must be generated on a project-
specific basis. (Refer to Chapter 12 for additional information.)
b. When an analytical result is being compared to a decision limit, it may be useful (e.g.,
for the purposes of data qualification) to estimate an upper or lower confidence limit for the re-
11-6
EM 200-1-10
30 Jun 05
sult. If there is significant analytical bias (i.e., the percent recovery for the LCS is statistically
different from 100%), the result can be corrected for bias prior to estimating confidence limits.
Since low bias is more common than high bias for environmental analyses (e.g., for extractable
organic compounds) and is more likely to adversely impact data quality than high bias, only low
bias will be addressed. Upper confidence limits (UCLs) will be approximated by correcting for
low bias and taking random error into account. The upper confidence limits will then be com-
pared to project action levels to qualify results. This strategy will constitute a relatively conser-
vative approach for risk-based applications.
c. If the percent recovery of a target analyte in the associated LCS is not too close to zero
(e.g., the percent recovery is least 2030%), precision is in control, then an upper confidence
limit for a laboratory result may be approximated using the following equation1:
UCL(C,%R,) = u(C,%R,) [ C / (%R / 100) ] (11-1)
d. The measured concentration of the sample and percent recovery for the associated
laboratory control sample are denoted by C and %R, respectively. The second term in Equation
11-1 (enclosed in brackets) is the biased corrected concentration. The first term, u(C,%R,),
will be referred to as the uncertainty factor because it accounts for the random error associated
with the measured result C and the calculated percent recovery %R. The factor is primarily a
function of C, %R, and the desired level of statistical confidence, . The factor will be some
positive value greater than one. The use of a high value for the uncertainty factor will result in a
conservative estimate for the UCL (e.g., will minimize false negatives when comparing results to
an AL).
e. If normality is assumed and the relative uncertainty (i.e., the relative standard devia-
tion) is assumed to be constant within the quantitation range of the method, then the uncertainty
factor for the 95% UCL may be estimated using the following equation: 2
u(95%) 1 + (2) (L95% / %R) (11-2)
where L95% is half the width of the warning range for the LCS percent recoveries (e.g., from the
laboratorys control charts). The half width of the control range, L99%, gives an upper 99% upper
confidence limit.
u(99%) 1 + (2) (L99% / %R) (11-3)
Note that the uncertainty increases as the width of the warning or control ranges increases and
the percent recovery decreases.
1
For a rigorous treatment of propagation of analytical measurement uncertainty, refer to the following reference:
Draft EURACHEM/CITAC Guide Quantifying Uncertainty in Analytical Measurement, Second Edition, June
1999, EURACHEM Measurement Uncertainty Working Group.
2
Georgian, T. Estimation of laboratory uncertainty using laboratory control samples. Environmental Testing and
Analysis, Vol. 9, No. 6, p. 20. November/December 2000.
11-7
EM 200-1-10
30 Jun 05
f. The assumption that the relative standard deviation is constant will be valid for sample
concentrations sufficiently near the spiking concentration for the LCS (typically the mid-calibra-
tion range) and will be appropriate when the standard deviation is approximately a linear (in-
creasing) function of concentration. Uncertainty is often proportional to analyte concentration
when the measurements are well above the detection limits. The above equations will probably
result in reasonable estimates when there is no appreciable matrix interference or sample hetero-
geneity, measurements are within the calibration range of the method, and the analyte levels are
near the LCS spiking concentration. Note that the variability associated with the heterogeneity
of the sample matrix is not taken into account because the total uncertainty is estimated from the
LCS, which is typically a clean matrix such as reagent water or purified sand.
g. The use of the mean LCS recovery (%R), rather than the use of a single LCS recovery,
%R, associated with a batch of samples, will generally result in a more reliable estimate of the
UCL. This is especially true when extreme low bias (e.g., %R < 20% or 30%) or high method
variability exists. Under these circumstances, bias correction should be performed using the
mean percent recovery. If the mean LCS recovery is available (e.g., at least 20 or 30 data points
were used to establish the laboratorys in-house statistical warning and control limits) and the
method is in statistical control, then substitute (%R) for %R in Equation 11-1 and use the fol-
lowing uncertainty factors:
u(95%) 1 + (L95% /%R) (11-4)
u(99%) 1 + (L99% /%R) (11-5)
h. Note that (when bias correction is performed) the use of the mean recovery decreases
the uncertainty (and the UCL) because the mean recovery is a more confident representation of
true bias than any single recovery value.
i. If there is no significant bias (i.e., %R = 100%), the relative uncertainty is approxi-

mately constant within the quantitative range of the method and the associated LCS recovery is
in control for the sample batch, then Equation 11-1 and either Equation 11-4 or Equation 11.5
may be used to estimate an upper confidence limit, by setting %R = 100:
UCL(95%) = u(95%) C (1 + L95% / 100) C (11-6)
UCL(99%) = u(99%) C (1 + L95% / 100) C (11-7)
j. Note that the total uncertainty is larger when a bias correction is performed. This oc-
curs because Equation 11-1 contains two sources of uncertainty (the uncertainty associated with
%R and C) while Equations 11-6 and 11-7 contain only one source of uncertainty (uncertainty
associated with C).
k. To illustrate the use of the above equations, assume that %R = 40% and C = 2 ppb. If
the LCS warning range is 60140%, then L95% = 40%. It follows from Equations 11-1 and 11-2
that the upper confidence limit for the measured result C is:
11-8
EM 200-1-10
30 Jun 05
UCL(95%) = (1 + 1.4) [ 2 ppb / (40% / 100) ] 12 ppb
l. If there is no significant method bias and the LCS recovery is in control, then the upper
confidence limit can be estimated using Equation 11-1 and Equation 11-6:
UCL(95%) = 1.4 (2 ppb) 3 ppb
m. Once an upper confidence limit is calculated, the upper confidence limits can be com-
pared to the project decision limits and this information can be used to qualify the data. To il-
lustrate, let %R = 40%, C = 2, and L95% = 40% (the first example presented above). Assume that
the project-required acceptance range for the LCS is 80120% and the project action level (AL)
is 50 ppb. Since the LCS recovery is 40%, the result C = 2 must be qualified (e.g., as estimated
or rejected). Since UCL(95%) = 12 ppb < AL = 50 ppb, despite the low bias, it is not likely that
the analyte is actually present in the sample at a concentration that exceeds the AL. Hence, it
would be appropriate to qualify the 2-ppb result with the J-, flag. However, if AL = 5 ppb, since
the UCL > AL, it may be more appropriate to qualify the result with the X flag (e.g., when
statistical analyses are not being performed and each reported sample concentration is being
directly compared to the AL). The low-biased result of 2 ppb does not demonstrate that the
analyte is present at a level that is less than the 5-ppb action level
n. It should be noted that the uncertainty factor does not typically exhibit a large amount
of variability in the context of the tolerances normally applied to laboratory environmental
analyses. The uncertainty factor will typically assume values between two to four, and, at worst,
will probably be less than ten. For example, if %R = 20% and the LCS control range is 20% to
180%, conditions that are indicative of rather poor method performance for a target analyte, then
an uncertainty factor of less than seven would be calculated from Equation 11-3. Therefore, if
the LCS recovery is unacceptably low but the recovery is not less than about 20%, then it may be
more convenient to calculate an UCL for a measured sample concentration by correcting the
measured concentration for bias and then simply multiplying the bias-corrected result by a factor
of five or ten. The UCL could then be compared to the AL to qualify a sample result associated
with the noncompliant LCS recovery. For example, if the UCL were less than the action, then
the result would be qualified as estimated (e.g., using the J- flag). If the UCL were greater than
the AL, then the sample result would be qualified potentially rejected (using the X flag).
11-9
EM 200-1-10
30 Jun 05
Table 11-2
Data Qualification for LCS Results 1
Acceptance Criteria: 80% %R 120%, RPD 20%

%R [RPD] Remarks [Bias] Sample (y) Sample Flag
MRL < MQL < y Flag not required.
MRL < y < MQL J
90% [18%] %R and RPD in
control y < MRL U
%R acceptable y > MRL J
90% [30%] RPD OFC
[Unknown] y < MRL UN
y > MRL J-
70% [15%] %R < LCL
[Low] y < MRL UN
y > MRL J+
140% [10%] %R > UCL
[High] y < MRL U
J-
y > MRL X if y < AL
%R << LCL
10% [15%] [Low] y < MRL R
J+
y > MRL Possibly X if y > AL
%R >> UCL
250% [20%] [High] y < MRL U
J
250% [200%] %R >> UCL
X if y < AL
or or %R << LCL
y > MRL Possibly X if y > AL
10% [200%] RPD grossly OFC
[Unknown] y < MRL R
Notes: 1. %R and RPD denote the percent recovery for the LCS and the relative percent difference for the
LCS/LCSD, respectively. The concentration of the field sample is denoted by y and the action level by AL. (It is
assumed that MRL < AL.) The terms out of control, upper control limit, and lower control limit are
abbreviated as OFC, UCL, and LCL, respectively. The inferred direction of bias is enclosed in brackets. The
symbols << and >> denote much less than and much greater than, respectively.
11-10
EM 200-1-10
30 Jun 05
CHAPTER 12
Matrix Spikes, Matrix Spike Duplicates, and Matrix Duplicates
12-1. Introduction.
a. Matrix spike (MS), matrix spike duplicate (MSD), and matrix duplicate (MD) re-
sults are examined to evaluate the impact of matrix effects on overall analytical performance and
the potential usability of the data. A matrix spike is a representative environmental sample that
is spiked with target analytes of interest prior to being taken through the entire analytical process
in order to evaluate analytical bias for an actual matrix. A matrix duplicate is a collocated (e.g.,
a VOC soil sample) or a homogenized sample that is processed through entire analytical proce-
dure in order to evaluate overall precision for an actual matrix. Duplicate or replicate matrix
spikes are also used to evaluate overall precision.
b. Matrix spike recovery failure and poor precision may arise because of (i) poor sam-
pling technique, (ii) inadequate homogenization, or (iii) from matrix effects associated with the
preparatory or determinative portion of an analytical method. For example, inappropriate sample
collection and handling procedures for VOC soil samples (e.g., as described in Method 5030)
may result in variable losses of VOCs, giving rise to poor precision and low bias. Sludges,
clayey soils or sediments, multi phasic samples, and samples with macroscopic particles of ana-
lytes such as explosives and metals, may defy homogenization attempts during sample prepara-
tion or compositing procedures used for sample collection, giving to unacceptable duplicate pre-
cision or matrix spike recoveries.
Note: In this document, sample heterogeneity arising from the spatial or temporal dis-
tribution of the analytes in a study area is viewed as a characteristic of the environ-
mental population being sampled and not as an interference that the method of analy-
ses must be optimized to address.
12-2. Interpretation of Matrix Spike and Duplicate Results.
a. In general, when evaluating accuracy using matrix spike recoveries, a matrix effect is
inferred when (i) all instrument and method QC samples (the LCSs and CCVs) are acceptable,
(ii) the spiking concentration for the matrix spike is high relative to the native analyte concen-
tration, and (iii) the recovery of the matrix spike does not fall within the laboratorys corre-
sponding statistical range for laboratory control samples. Similar reasoning applies to the
evaluation of precision using RPDs for MS/MSDs and MDs results. Namely, an interference is
inferred when (i) instrument and method QC is in control, (ii) the native analyte concentrations
are sufficiently high (e.g., above the quantitation limits), and (iii) some measure of precision
(such as the RPD) exceeds the corresponding statistical LCS limits.
b. Laboratory and project documents (e.g., laboratory standard operating procedures and
QAPPs) often state that the presence or absence of matrix effects is determined by establishing
statistical control ranges using MS rather than LCS spike recovery data. Once the MS control
limits are established, a matrix effect is subsequently inferred for a batch of environmental sam-
12-1
EM 200-1-10
30 Jun 05
ples if an associated matrix spike recovery falls outside of the statistical MS control range (rather
than outside of the LCS control range). This approach will typically be inappropriate! In order
for this strategy to be viable, the matrix used to establish the MS control range must be relatively
uniform, similar in composition to the environmental matrix of interest, and known to lack sig-
nificant interferences.
c. Because of the variety and complexity of environmental matrices, it is usually

impractical for environmental production laboratories to establish matrix-specific control limits.
Most (if not virtually all) environmental laboratories that maintain statistical MS control ranges,
establish MS limits by method rather than by matrix. For example, groundwater, surface water,
rain water, and waste water are often erroneously considered to be the sample matrix for the
purpose of calculating statistical MS control limits because the samples are processed using the
same aqueous preparatory and determinative methods. Furthermore, the MS control ranges are
frequently calculated using MS recoveries that have been impacted by matrix effects. These
problems frequently result in very wide MS control limits that are difficult to interpret and
frequently do not satisfy project objectives. Furthermore, since the MS control ranges are often
calculated using spiked samples affected by significant matrix inferences, the absence of a
matrix effect is not demonstrated when a MS recovery for a batch of environmental samples falls
within the MS recovery range. At best, the result may demonstrate that a matrix effect (if
present) is no larger than is typically observed for a variety of matrices analyzed by the same
preparatory and determinative method.
d. In general, matrix spike control limits are not available from environmental
production laboratories as off-the-shelf commodities but must be established on a project-
specific basis. In order to obtain representative matrix spike control limits, a relatively large
number of matrix spike samples (e.g., 20 to 30 samples) must be taken from each environmental
medium in each project study area. When a projects matrix spike acceptance ranges are
established solely upon the basis of a laboratorys statistical MS control limits and these limits
were developed using MS recoveries from non-project related media or dissimilar matrices that
have been impacted by interferences, then the matrix spike control limits will probably be
inappropriate. Before proceeding with the data evaluation, assess the validity of the matrix
spike acceptance limits (e.g., determine whether the acceptance ranges are unrepresentative or
too wide to satisfy projects data objectives). A strategy for approximating statistical matrix
spike control ranges using LCS recovery data is presented in Paragraph 12-3
e. Lastly, it should be noted that matrix spike recoveries are evaluated, at least
potentially, to fulfill two separate objectives: (i) To determine whether or not matrix effects exist
and (ii) to determine whether or not project-specific objectives for accuracy were satisfied for
the analytes in the matrices of interest. The distinction between the two objectives is somewhat
subtle but important to recognize when qualifying data because data are frequently qualified
(e.g., as estimated) on the basis of the second objective rather than the first.
f. To illustrate the evaluation of matrix spike and LCS results, assume that a laboratorys
statistical control range for LCS recoveries for aqueous lead analyses is 80120%, the project-
required acceptance range for MS recoveries is 50150%, and three separate sets (batches) of
samples were analyzed with associated MS recoveries of 90, 65, and 40%. Assume that the
12-2
EM 200-1-10
30 Jun 05
spiking concentrations for all three MS samples are high relative to the native analyte concentra-
tions and QC is otherwise acceptable. Since the 90% MS recovery lies within the statistical LCS
acceptance limits, this recovery suggests the absence of any matrix effects. Since the MS recov-
ery of 65% falls well outside of the LCS statistical acceptance range, the recovery is indicative
of a matrix effect that is within the project-required tolerance for accuracy (50150%). Al-
though the recovery is indicative of matrix interference, data qualification would not necessarily
be required. The recovery of 40% is indicative of a matrix effect that is greater than the project-
required tolerance for matrix effects. At a minimum, data qualification would typically be re-
quired.
12-3. Estimating Statistical Matrix Spike Recovery Ranges.
a. If the spiking concentration for the MS is at least twice as large as the native analyte
concentration, the laboratorys in-house statistical control or warning limits for LCS recoveries
can be used to establish acceptance limits for MS recoveries:
%R L95% (100 /%R) (%R / 100 + CB / CS) (12-1)
%R L99% (100 /%R) (%R / 100 + CB / CS) (12-2)
b. As defined in Chapter 11-6, [%R] is the mean LCS recovery, L95% is the half width of
the LCS warning range and L99% is half the width of the control range. The variable CB denotes
the native analyte concentration (i.e., the measured pre-spike sample concentration) and CS de-
notes the calculated spike concentration in the sample matrix (i.e., the analyte concentration
added to the sample matrix). If method bias is not significant (i.e., [%R] is near 100%), then the
following equations may be used to estimate the MS acceptance ranges:
%R L95% (1 + CB / CS) (12-3)
%R L99% (1 + CB / CS) (12-4)
c. For example, if the LCS acceptance range is 80120% (i.e., 100% 20%) and the
spike concentration is twice the native analyte concentration, then the acceptance range for the
MS recovery is as follows:
100 20% (1 + ) = 100 30% = 70130%
d. Therefore (in this example), if the LCS recovery for a batch of environmental samples
falls within 80120% but the recovery of the associated matrix spike does not fall within 70
130%, then a matrix effect would be demonstrated.
e. The acceptance range for MS recoveries may be set equal to the acceptance range for
LCS recoveries when the MS spike concentration is much higher than the native analyte con-
centration (e.g., by a factor of five to ten) or when it is desirable to establish a conservative (i.e.,
a more narrow) MS acceptance range.
12-3
EM 200-1-10
30 Jun 05
Note: Since two measurements are required to calculate a MS recovery (the pre-spike
and post-spike sample concentrations) but only one measurement is required to cal-
culate the LCS recovery (the post-spike sample concentration), in order to establish
MS acceptance limits from the statistical LCS acceptance limits, the random error asso-
ciated with the additional MS measurement must be taken into account. (A pre-spike
sample concentration is not measured for the LCS; since the LCS is a spiked blank, the
pre-spike sample concentration is assumed to be zero.) The correction factors en-
closed in parentheses in Equations 12.1 to 12.4 account for the additional measurement
uncertainty associated with MS recovery determinations. The correction factors were
calculated by assuming that the standard deviation is a linear function of concentration
and give first-order approximations for the MS acceptance limits.
12-4. Criteria.
12.4.1. Representativeness.
a. Before evaluating matrix spike results, review the SAP, QAPP and similar planning
documents. These documents should describe how representative matrix spikes will be selected
for the environmental matrices of interest, particularly for heterogeneous matrices such as soils.
b. The composition of a matrix spike sample must be similar to that of the associated
environmental samples. For example, when soil sampling is performed, the SAP should describe
how the on-site geologist will select representative matrix spikes. This typically entails classifi-
cation of soil type. For example, a matrix spike should be collected for a set of samples high in
sand and a separate matrix spike should be collected for a set of samples high in clay. However,
this does not imply that matrix spikes should be collected solely on the basis of grain size classi-
fication (e.g., sand, silt, and clay). For example, the origin of the geological formation (fill, gla-
cial deposits, stream deposits, etc.) should also be taken into account. Therefore, unless all soil
samples are being collected in a single geological formation of relatively uniform composition or
matrix interference has been well characterized during prior investigations, a batch of samples
should typically contain several matrix spikes (each representing a different soil type and general
origin). Similarly, if only one matrix spike were collected for a set of groundwater samples but
the groundwater samples were collected from two hydraulically isolated aquifers being investi-
gated at the site (e.g., a shallow and a deep aquifer), then, in general, one should not assume
that the matrix spike would be representative of the groundwater in both aquifers.
c. If the matrix spike sample for the preparation batch originates from a different project
site or is suspected to be of dissimilar composition from the other samples in batch, it must not
be used to qualify the other field samples. In order to consolidate small numbers of samples
from different project sites, the laboratory may analyze samples from different projects together
in the same preparation batch for the same parameters. However, the MS results would not be
applicable to the samples collected from the other sites. Allowing the laboratory to choose the
samples to be spiked often results in the selection of unrepresentative matrix samples. Similarly,
matrix spikes must not be selected by field personnel in a manner that is solely designed to sat-
isfy frequency requirements. For example, the collection of all matrix spikes on the last day of
12-4
EM 200-1-10
30 Jun 05
sampling activities to satisfy a 5% frequency requirement for the collection of matrix spikes will
typically result in unrepresentative matrix spike samples.
d. In general, a matrix spike sample must contain all the target analytes of interest. A
subset may be used when it can be demonstrated that the subset of target analytes characterizes
(i.e., represents) method performance for the remaining (unspiked) target analytes.
Note: When only a subset of the target analytes is included in the matrix spikes, project
documents such as the QAPP must present a scientifically defensible rationale for not
spiking the entire set of target analytes. A number of promulgated analytical methods
recommend specific target analytes for the matrix spikes. Merely referencing a subset
of analytes recommended in a published analytical method (e.g., the six MS com-
pounds listed in SW-846 Method 8260B) does not constitute a scientifically defensible
rationale for not spiking all the target analytes (e.g., unless the method explains why the
subset of spiked analytes is representative of the remaining target analytes).
12-4.2. Frequency.
Review the appropriate project documents (e.g., the QAPP) to determine the required frequency
of MSs, MSDs, and MDs. A MS and MSD or MS and MD (representative of each type of ma-
trix analyzed) are usually required for every batch of samples processed. MD pairs are typically
used for inorganics (especially metals) and MS/MSDs for organics. Matrix spikes and matrix
duplicates are usually collected at a frequency of at least 5% if the matrix is relatively uniform in
physical composition.
Bias and precision specifications for matrix spikes and matrix duplicates are dependent upon the
DQOs of the investigation. Acceptance limits for matrix spikes and duplicates should be speci-
fied in project documents such as the QAPP. Guidance for establishing default acceptance
limits for matrix spikes and matrix duplicates (e.g., when acceptance limits are not specified) is
presented below.
12-4.3.1. Project Specific Communications.
a. The laboratorys statistical LCS acceptance limits should not be greater than the pro-
ject-required acceptance limits for matrix spikes and matrix-dependent duplicates. When this
criterion is not satisfied (i.e., project-required acceptance limits are more stringent than the sta-
tistical LCS acceptance limits) and matrix spikes or matrix-dependent duplicates fail to meet the
project-required acceptance limits, it is not generally valid to assume that the failures resulted
from matrix effects. For example, assume that the statistical LCS recovery range is 60140%,
the project-required MS recovery range is 80120%, and a MS recovery for a batch of environ-
mental samples is 65%. The associated environmental samples must be qualified (e.g., using the
J flag) for not meeting the project-required tolerance for accuracy. However, the associated
sample results must not be qualified for matrix interference. (In this example, the MS recovery
of 65% falls well within the statistical LCS acceptance range).
12-5
EM 200-1-10
30 Jun 05
b. If the acceptance limits for matrix spikes are not specified or are inappropriate (e.g.,
refer to Paragraph 12-2) and the laboratorys statistical LCS acceptance ranges are comparable
to or more stringent than the project-required LCS acceptance ranges (e.g., the warning or
control ranges for the LCS recoveries fall approximately within the corresponding project-
required acceptance ranges for the LCS recoveries), then approximate the statistical matrix spike
recovery ranges as discussed in Chapter 12-3. Compare the calculated MS acceptance ranges
and the project-required LCS acceptance ranges. Qualify the environmental data using the most
extreme limits from the two sets of acceptance ranges. However, it is emphasized that this
approach is applicable only if the project-required LCS ranges are greater than or equal to the
laboratorys statistical control ranges.
Note: It is recommended that the ranges be rounded (e.g., to the nearest 5% or 10%) to
more readily compare the laboratorys statistical acceptance range to a project-
required acceptance range. It is also recommended that the laboratorys statistical
limits be viewed to be comparable to the project limits, when the LCS warning range
falls approximately within the project-required LCS acceptance range. Alternatively,
the width of the control range should be no greater than about 1.5 times the projects
acceptance range. For example, if the project-required recovery range for the LCS is
90% - 110%, then a warning range of 90% - 110% or a control range of 85% - 115%
would be considered to be acceptable.
c. To illustrate the above approach, assume that a matrix spike acceptance range is not
specified, the laboratorys statistical control range for the LCS is 67113% (i.e., 90% 23) and
the project-required acceptance range for the LCS is 70130%. The laboratorys statistical con-
trol range approximately falls within project-required LCS acceptance range. If it is assumed
that the spiking concentration for the MS is at least twice as large as the native analyte concen-
tration (e.g., which will typically result in a conservative estimate for the MS acceptance range),
then, using Equation 12.4 in Chapter 12.3. In this example, the acceptance range for the MS is
90% 23% (1.5) = 55125%. (Note that if Equation 12-2 were used, the acceptance range
would be only be slightly wider: 90% 23% (1.6) = 53127%.) Since the calculated MS ac-
ceptance range is 55125% and the project-required LCS acceptance range is 70130%, set the
MS acceptance range for the project using the most extreme limits; use 55130% as the MS ac-
ceptance range. Therefore, a MS recovery that does not fall within 65130% is indicative of a
significant matrix effect and the associated environmental samples would be qualified (e.g., as
estimated or potentially rejected).
d. When acceptance limits for matrix spikes recoveries are not specified or are
inappropriate and the laboratorys statistical LCS control ranges are significantly wider than the
project-required LCS acceptance ranges, then a conservative approach is recommended. Evalu-
ate the matrix spike recoveries using the project-required LCS acceptance limits. For example,
if the LCS acceptance range is 80% - 120%, then the matrix spike acceptance range should be set
to 80% - 120%. If LCS acceptance limits are not specified, then use the guidance presented in
Chapter 11 of this document to establish a set of default LCS/MS acceptance limits. In gen-
eral, if the MS recovery falls outside of the LCS acceptance range, then qualify the associated
results as estimated or rejected. However, it is inappropriate to attribute the unacceptable MS
12-6
EM 200-1-10
30 Jun 05
recovery solely to matrix interference. The evaluation strategies for the matrix spike and matrix
duplicates are essentially the same as those for laboratory control samples described in Chapter
11.
12-4.3.2. Establishing Acceptance Limits for Matrix-Dependent Duplicates.
If acceptance limits are not specified for matrix-dependent duplicates (i.e., MDs and MS/MSD
pairs), if appropriate, then calculate the matrix spike limits using the procedure Chapter 12.3 and
set the maximum RPD equal to one half the calculated MS acceptance range. Alternatively,
evaluate the RPD results using the project-required RPD acceptance limits for laboratory control
samples. If RPD limits are not specified for laboratory control samples, set each RPD accep-
tance limit for matrix-dependent duplicates equal to one half of the width of the project-required
recovery range for the corresponding LCS, or to the laboratorys statistical RPD acceptance limit
when derived from LCS data, whichever is less. For example, if the project-required LCS re-
covery range is 80% - 120% and the laboratory does not maintain statistical limits for duplicate
precision using LCS data, set the RPD acceptance limit for matrix-dependent duplicates to 20%.
12-5. Evaluation.
Review the standard preparation logs to verify that all target analytes were included in the matrix
spike. Using the laboratory summary forms for the matrix spike and matrix duplicate results,
recalculate the recovery and the RPD for at least one target analyte. Compare the calculated val-
ues to the values reported on the laboratorys summary form. The result must agree to within
two significant figures. Review the Case Narrative and all of the recovery and precision results
on the laboratory summary forms and note any failures.
a. Contractual issues may impact the review of MS, MSD, and MD data. However,
contractual considerations for matrix spikes and matrix duplicates are more complex than those
for blanks and laboratory control samples because the results are dependent upon matrix effects
as well as sample preparation and analysis errors. For example, the heterogeneity of soil grab
samples and sequentially collected groundwater samples complicates the evaluation of MS/MSD
results because uniform concentrations are assumed for the native analytes. Therefore, laborato-
ries do not typically base batch control on the results of MS, MD, or MSD samples unless a gen-
eral method failure is indicated.
b. When matrix spikes or matrix duplicates grossly fail QC acceptance limits in a

systematic manner, examine the Case Narrative and any laboratory communications (e.g., phone
logs) included in the data package to determine if the Project Manger was notified and corrective
actions other than data qualification were performed. Refer to project planning documents such
as the Scope of Work for laboratory analytical services and the QAPP to determine whether cor-
rective actions other than data qualification are required.
c. When gross failures occur and expected laboratory corrective actions are not per-
formed, the reviewer should consult with the Project Manger to determine whether to proceed
12-7
EM 200-1-10
30 Jun 05
with the PB review or to reject the data package as a whole (e.g., the laboratory may be required
to reanalyze the environmental samples). Some probable corrective actions for matrix interfer-
ences are listed below:
(1) If a matrix spike recovery is unacceptable and matrix interference is suspected, then
the laboratory should be expected to make a reasonable attempt to remedy the problem.
Corrective action for matrix interference may include the implementation of cleanup procedures
or other method modifications. For example, cleanup methods should be performed to address
matrix interferences for extractable organic analyses such as the BNA, pesticides, and PCB
analyses (e.g., as described in SW-846 Method 3600). The method of standard additions may be
required for metal analyses. Under these circumstances, verify that appropriate method
modifications were performed to minimize the matrix interference.
(2) When a MS recovery is unacceptable but matrix interference is not otherwise appar-
ent, the MS sample would normally be reprocessed (e.g., reextracted and reanalyzed) by the
laboratory to verify the effect. However, the MS sample would not be reprocessed if the failure
is consistent with historical data. The matrix effect is confirmed if the second result is similar to
the original result (in magnitude and direction of bias). It should be noted that some methods
specify other verification procedures. For example, if low matrix spike recoveries are obtained
for hexavalent chromium in soil, Method 3060A indicates that additional analyses should be
performed (e.g., pH and oxidation-reduction potential) to determine whether or not the low ma-
trix spike recovery results from reducing conditions within the environmental sample. When un-
acceptable matrix spike recoveries are obtained, examine the data package to determine if appro-
priate confirmatory procedures were implemented.
Data that fail quality objectives because of matrix effects may be unusable to support decisions
and must be qualified. Data quality may also be adversely impacted if the matrix spike sample is
not representative of the other environmental samples in the batch. Data are qualified for matrix
effects primarily using the same qualification strategies for laboratory control samples. In par-
ticular, data qualification must take both magnitude and direction of bias into account. When
both a MS and MSD are processed for a batch of samples, use the most noncompliant matrix
spike recovery to evaluate and qualify the data. Additional guidance is presented below.
12-7.1. Matrix Spikes and Matrix Spike Duplicates.
a. For both the MS and MSD, compare the spiking levels to the concentrations of the na-
tive analytes in the sample selected for spiking. If the native concentration of a target analyte is
high relative to the spiking concentration, then this may contribute a significant uncertainty to
the recovery calculations; the MS recovery may not be representative of actual method perform-
ance for the matrix. In the absence of other guidance, evaluate the MS recovery when the spik-
ing concentration is at least two times greater than the native analyte concentration. If envi-
ronmental samples were qualified by the laboratory for matrix interference but the spiking levels
are low relative to the native analyte concentrations, then the flags must be omitted. However,
professional judgment is important when evaluating the native analyte concentration relative to
12-8
EM 200-1-10
30 Jun 05
the spiking concentration. For example, if the spiking concentration is near but less than two
times the native analyte concentration, a gross MS recovery failure (e.g., a MS recovery of 5%)
is probably indicative of a matrix effect (rather than a low-spiking concentration) and the associ-
ated results must be qualified for matrix interference. In general, if the MS spiking concentra-
tion is between one and two the native analyte concentration, then data qualification is recom-
mended only when gross MS recovery failures occur.
b. If the LCS results are acceptable, the spiking levels for the MS are high relative to the
native analyte concentrations (i.e., at least two times the native analyte concentration), the matrix
spike sample is representative of the other environmental samples, and the MS recovery falls
outside of the acceptance limits, then significant matrix interference may exist. Qualify the as-
sociated sample results (e.g., environmental samples of a similar matrix collected from the same
site) as follows:
(1) If all target analytes are present in the matrix spike, and the recovery of a particular
analyte is unacceptable, then qualify all detections of the analyte in the associated environmental
samples using the strategies discussed in Chapter 11. For example, if the MS recovery for a tar-
get analyte falls grossly below the lower recovery acceptance limit, then qualify all detections
less than the AL with the X flag. Note that in those instances where it can be determined that the
MS or MSD results affect only the sample spiked, qualification must be limited to this sample
alone.
(2) If all the target analytes are not present in the matrix spike, then use professional
judgment to determine the extent to which qualification of the non-spiked target analytes is
required. In general, each spiked analyte must be clearly linked to each of the unspiked target
analytes. If one of the spiked analytes clearly represents some subset of the target analytes, then
qualify only the target analytes of the subset on the basis of the MS recovery. For example, if
analyte A in the matrix spike sample is representative of the subset of target analytes {A, B,
C} in the environmental samples, then qualify analytes A, B, and C for the environmental
samples using the MS recovery of analyte A. However, if a clear association does not exist
(e.g., and the lack of matrix interference was not demonstrated during a prior sampling event),
then a conservative approach is recommended. At a minimum, qualify detections and
nondetections for the unspiked analytes in the environmental samples as estimated (i.e., qualify
detections with the J flag and nondetections with the UN flag). However, if the recovery of one
or more of the spiked analytes is unacceptable, then qualify all of the unspiked analytes using the
most noncompliant MS recovery.
c. If a MS sample is not available or is not representative of the other samples in the

batch, then the performance of the method in the matrix of concern has not been well character-
ized. At a minimum, qualify the environmental sample results as estimated. If the data are being
used to support critical decisions and method performance in the matrix of concern is not other-
wise known (e.g., the environmental population of interest has not been previously sampled and
the surrogate recoveries are not available or representative of the target analyte), then it may be
appropriate to qualify the sample results as tentatively rejected.
12-9
EM 200-1-10
30 Jun 05
12-7.2. Matrix-Dependent Duplicates.
a. Precision is typically measured using the RPDs for MS/MSD or MD pairs. MS/MSD
pairs would normally be used to evaluate duplicate precision when low-level contamination is
anticipated (i.e., analyte concentrations less than the MQLs) and MDs would normally be used to
evaluate duplicate precision when high levels of contamination are expected. Compare the
RPDs reported for all target analytes to the corresponding RPD acceptance limits.
b. Evaluate target analyte RPDs for MS/MSD pairs when the spike concentration is at
least two times the native analyte concentration. Evaluate target analyte RPDs for MD pairs for
analytes detected at or above the MQL. (The RPD is evaluated when a target analyte detection is
greater than or equal to the MQL for at least one sample of the MD pair.) RPD results that do
not satisfy these criteria (e.g., RPDs calculated from detections at concentrations less than the
MQLs) must not be used to evaluate duplicate precision.
Note: Sometimes an acceptance criterion for duplicate precision is specified for the
MQL and a different acceptance criterion is specified for concentrations that are greater
than the MQL by some multiplicative factor. Evaluate the appropriateness of the du-
plicate precision acceptance criterion that is nearest to the decision limit prior to per-
forming data qualification. For example, assume that the QAPP requires the maximum
RPD to be 40% for results equal to or greater than five times the MQL and requires re-
sults to agree to within MQL for concentrations between the MQL and 5 x MQL.
Also assume that AL = 32 ppb, MQL = 20 ppb, and the following duplicate results are
obtained: 20 ppb and 40 ppb. Since the duplicate results are less than 5 x MQL (100
ppb) and agree within MQL (i.e., 20 ppb), according to the QAPP, the results
should not be qualified. However, since the MQL is near the AL and the RPD for the
duplicate pair is high (RPD = 67%), the duplicate results do not demonstrate that con-
tamination is above or below the AL. Contrary, to the criteria specified in the QAPP,
qualified the associated sample results as estimated (e.g., unless quantitative statistical
methods are being used to quantify the uncertainty and to compare the results to the
AL).
c. If (i) the LCS results are acceptable, (ii) the spiking levels for the MS/MSD are high
relative to the native analyte concentrations (i.e., at least two times the native analyte concentra-
tion) or the native analyte concentrations for the sample/MD are at least as high as the MQL, and
(iii) the RPD is unacceptable, then a significant matrix effect may exist.
d. If precision is evaluated using MS/MSD pairs containing only a subset of the target
analytes of interest and the analytes are representative of the set of unspiked target analytes,
then qualify the sample results using the subset of target analytes in the MS/MSD. If it is un-
known whether or not the subset of target analytes adequately represents the unspiked target
analytes, then a conservative approach is recommended. Evaluate the unspiked target analytes
using the most noncompliant RPD for the MS/MSD. However, even when duplicate precision is
acceptable for the subset of target analytes in the MS/MSD, it may be appropriate to qualify all
detections and nondetections of the unspiked target analytes as estimated (e.g., when statistical
12-10
EM 200-1-10
30 Jun 05
analyses are not being performed to characterize the variability of these analytes in the matrix of
concern).
e. When the RPD is unacceptable, qualify the associated sample results using the same
strategies presented in Chapter 11 (e.g., Table 11-2). For example, when precision is evaluated
using MD pairs or MS/MSD pairs and the direction of bias is unknown, then qualify all detec-
tions of the analyte in the associated environmental samples with the J flag and nondetections
with the UN flag when marginal failures occur. However, when the RPD is marginally unac-
ceptable and the direction of bias can be determined from other QC information, then qualify the
detections using J+ or J- flag (instead of the J flag). For example, assume that the acceptance
range for matrix spike recoveries is 80120%, the acceptance limit for the RPD is 20%, and an
RPD of 33% was calculated from matrix spike recoveries of 70 and 50%. Since the RPD is mar-
ginally unacceptable and bias is low, the associated detections would be qualified with the J-
flag. However, in those instances where it can be determined that the results affect only the MD
or MS/MSD pairs (and not the other samples in the preparation batch), then qualification must be
limited to those samples alone.
f. It may not be possible to collect representative duplicates. For example, if duplicates

are collocated samples (e.g., a pair of VOC soil samples) or cannot be homogenized because of
the nature of material being sampled (e.g., multi phase wastes), then high RPDs are probably the
result of sample heterogeneity rather than method performance problems in the matrix being in-
vestigated (e.g., digestates with high concentrations of dissolved salts, being analyzed for trace
metals by Method 6010B, are not intermittently clogging the ICP nebulizer, giving rise to erratic
results). If precision failures occur (gross or marginal) sample heterogeneity, then it is recom-
mended that detections be qualified with the J flag and nondetections be qualified with the UN
flag. The data review report must state that representative duplicates were not collected and the
data user should determine whether or not the environmental sample and matrix-dependent du-
plicate results can be used to support project decisions.
12-11
EM 200-1-10
30 Jun 05
CHAPTER 13
Surrogates
13-1. Introduction.
Surrogates are organic compounds that are similar in chemical composition to the analytes of
interest and spiked into environmental and batch QC samples prior to sample preparation and
analysis. Surrogate recoveries for environmental samples are used to evaluate matrix
interference on a sample-specific basis. However, in order for this approach to be viable, the
surrogates must behave in the same manner as the corresponding target analytes that are native
to the matrices of interest (e.g., must partition between various phases in the same manner as the
native target analytes). Unfortunately, in practice, this equivalency is typically difficult to
demonstrate and is often more assumed than empirically derived. The most representative
surrogate will typically be an isotopically-modified version of the target analyte. Therefore,
when evaluating surrogate results, the representativeness of the surrogates should always be
taken into account.
13-2. Criteria.
a. The acceptance for surrogate recoveries must take the end use of the data into
account and must not be based solely upon contractual or method-specified limits. Method-
specified surrogate acceptance limits (e.g., for SW-846 and CLP methods) are often
inappropriately wide. Statistically-based acceptance limits generated by the laboratory may be
representative of routine method performance but may also be too wide (i.e., may not satisfy
project-specific DQOs).
b. The acceptance ranges for surrogate and target analyte spike recoveries must be
similar (particularly for laboratory control samples and blanks), since, by definition, surrogates
and target analytes are chemically similar compounds.
Note: It is common for statistical control limits for surrogates to be significantly wider
than the control limits for target analytes. This often occurs when surrogate control
limits are calculated by inappropriately grouping surrogate recoveries from LCSs, MSs,
and environmental samples into a single data set.
c. When the surrogate acceptance ranges are significantly wider than the acceptance
ranges for the target analyte, then the appropriateness of the surrogate acceptance ranges must be
carefully evaluated prior to performing data review or validation. When the surrogate accep-
tance limits are inappropriately wide, establish default acceptance limits using the target ana-
lyte acceptance ranges if these ranges appear to be reasonable. For example, if the acceptance
ranges for the target analytes are approximately 70130% (e.g., for the LCS) and the surrogate
acceptance limits are 20150%, set the acceptance range for the surrogates to 70130%. Other-
wise (i.e., in the absence of more appropriate acceptance limits), surrogate recoveries for organic
methods should be evaluated using the acceptance ranges of 80120% for purge-and-trap meth-
ods and 60140% for extractable organic methods. However, if the LCS is prepared from an in-
dependent-source standard, then an acceptance range of 70130% may be used for purge-and-
trap methods.
13-1
EM 200-1-10
30 Jun 05
d. If an analytical method requires no more than two surrogates, then surrogate results
are acceptable only if all of the surrogate recoveries are in control. If three or more surrogates
are associated with a set of target analytes, then one surrogate may be marginally (but not
grossly) out of control. However, the marginal failure must not be systematic in nature (i.e.,
must occur in a sporadic or random manner). In particular, if several consecutive failures are
observed for the same surrogate, then the data must be qualified.
13-3. Evaluation.
Review the laboratory Case Narrative and the summary forms and note any surrogate failures
that are reported. A significant amount of professional judgement is required to evaluate
surrogate results. However, the following strategies are generally applicable:
a. Prior to reviewing the surrogate data, examine the Case Narrative to determine
whether any of the surrogate results should not be used to qualify the environmental sample
results.
(1) Do not qualify environmental samples for matrix interference when surrogate
recoveries are unacceptable because of localized chromatographic problems. For example, if
several surrogates are associated with a group of target analytes and some (but not all) of the
surrogate recoveries are unacceptable because of coeluting interferences, then qualification is not
required.
(2) Do not qualify environmental samples for matrix interference when surrogate
recoveries are unacceptable because of dilutions. For example, if all of the surrogate recoveries
for an environmental sample are unacceptable because the surrogates were diluted out, but the
surrogate recoveries for the LCS and associated blanks are acceptable, then no further action is
typically required.
(3) It is recommended that the raw data be requested for review when zero-percent
surrogate recoveries are reported and these recoveries are not attributed to dilution. Zero-percent
recoveries may arise from retention time shifts rather than from losses (e.g., during extraction).
b. If an unacceptable surrogate recovery is associated with only a subset of the target

analytes (e.g., the surrogate is representative of the performance for only the acid fraction of the
BNAs analyzed by Method 8270B), then qualify the results for only the subset of analytes.
c. Surrogate recoveries for laboratory control samples and method blanks characterize
overall laboratory method performance in the absence of matrix interference are evaluated in
much the same manner as target analyte recoveries. Distinguish unacceptable surrogate
recoveries arising from matrix effects beyond the control of the laboratory from failures arising
from poor laboratory analytical technique. When a surrogate recovery is out-of-control for an
environmental sample but is also out-of-control for the LCS or an associated blank (e.g., the
method blank), a laboratory performance problem rather than a matrix effect must be assumed.
13-2
EM 200-1-10
30 Jun 05
d. Check for transcription and calculation errors for a representative number of samples.
Using the laboratory summary form for the surrogate results, recalculate the recovery of at least
one surrogate and compare the calculated value to the reported value. The two results must
agree to within two significant figures.
a. Contractual considerations may impact the data review when surrogate failures are
observed for laboratory control samples and blanks. A laboratory would normally be expected
to reprocess a batch of field samples when a surrogate recovery is unacceptable for an LCS or
blank. When surrogate recoveries for laboratory control samples or blanks are unacceptable and
the batch of samples is not reprocessed, examine the Case Narrative and note why the corrective
action was not performed. When surrogate recoveries for laboratory control samples and blanks
grossly and systematically fail QC acceptance criteria, qualify the affected data accordingly and
notify the Project Manager to determine whether to continue the PB data evaluation. (If the
review were discontinued under these circumstances, the entire data package would be rejected.)
b. When surrogate failures are noted for environmental samples, refer to project
documents such as the QAPP and the Scope of Work for analytical services to determine what
corrective actions need to be documented in the laboratorys data package. Corrective actions
typically performed for surrogate failures are discussed below:
(1) If matrix interference is not apparent in the chromatogram, an unacceptable surrogate

recovery for an environmental sample is normally confirmed by reextracting and reanalyzing the
sample. (The extract would be reanalyzed for confirmation if there were insufficient sample for
reextraction.) The matrix effect is confirmed when the repeated result is within the same order
of magnitude and exhibits bias in the same direction as the original result. Under these circum-
stances, examine the data package to determine if confirmatory analyses were performed. How-
ever, it should be noted that the laboratory may not routinely reprocess environmental samples
with unacceptable surrogate recoveries unless surrogate failures in method blanks or laboratory
control samples are indicative of a general method failure.
(2) When surrogate recoveries are unacceptable because of matrix interference, the
laboratory may be required to perform method modifications or cleanup procedures (e.g., as
described in Method 3600 of SW-846 for the SVOC analyses). Under these circumstances,
examine the data package to determine if cleanups were performed. Note that when there are
unacceptable surrogate recoveries followed by successful reanalyses, the laboratory is typically
required to report only the successful run. When there are unacceptable surrogate recoveries
followed by unsuccessful reanalyses, the laboratory is typically required to report both runs.
13-3
EM 200-1-10
30 Jun 05
a. The qualification protocols for surrogate recoveries are similar to those for LCS
recoveries. Qualification is generally required when the surrogate acceptance criteria of Chapter
13.2 are not met. If two surrogates are associated with (i.e., are representative of the
performance of) a set of target analytes and both surrogate recoveries are unacceptable, qualify
the sample result using the most noncompliant surrogate recovery. Similarly, if three or more
surrogates are used and one or more surrogates are grossly out of control, then data qualification
must be based upon the most noncompliant surrogate recovery. However, no action is required
if three or more surrogates are used and one surrogate is marginally out of control in a sporadic
manner.
b. Data qualification for noncompliant surrogate recoveries is dependent upon the

direction and magnitude of the failure. Distinguish gross surrogate recovery failures from
marginal failures. In the absence of more appropriate guidance, a gross failure is defined to
occur when any surrogate recovery does not fall within 20180% for extractable organic
analyses and 60140% for purge-and-trap analyses.
c. When a surrogate recovery for an environmental sample falls outside of the

acceptance limits, the direction of bias will be said to be well defined when the remaining
surrogates and all associated QC samples are in control or exhibit bias in the same direction. For
example, if the recovery of a surrogate exceeds the upper control limit but the recoveries of other
surrogates are below the lower control limit, then the direction of bias is not well defined (i.e.,
has not been adequately demonstrated). When there are several surrogates, a high or low
recovery for a single surrogate is not necessarily indicative of the direction of bias or method
extraction efficiency.
d. A direction of bias must not be inferred from the surrogate recoveries of volatiles
analyzed by purge-and-trap (e.g., when the recoveries of all the surrogates are unacceptably low
or high) unless the responses of the internal standards are available for review. Similar
compounds are used for internal standards and surrogates for purge-and-trap analyses. The
direction of bias will not be well defined when the surrogate and internal standard recoveries are
not consistent with one another. For example, a high surrogate recovery can be obtained when
the internal standard response (e.g., peak area) is extremely low (since the concentration of the
surrogate is determined from the ratio of the surrogate response to the internal standard
response).
e. In general, when the criteria of Chapter 13.2 are not met, qualify the target analytes
(associated with the surrogate) as discussed below. The qualification strategies below apply (i)
when two surrogates are used, (ii) and when three or more surrogates are used and gross or
systematic surrogate failures are observed. These qualification strategies are illustrated in Table
13-1.
(1) If any surrogate recovery is marginally unacceptable, bias is well defined, and there
are no gross recovery failures for other associated surrogates, then the data must be qualified as
13-4
EM 200-1-10
30 Jun 05
follows: For low bias, qualify detections with the J- flag and nondetections with the UN flag.
For high bias, qualify detections with the J+ flag and nondetections with the U flag. (Note that
qualification is not required when three or more surrogates are used and one sporadic marginal
failure is observed.)
(2) If any surrogate recovery is marginally unacceptable, bias is not well defined, and
there are no gross recovery failures for other associated surrogates, then the data must be
qualified as follows: Qualify detections with the J flag and nondetections with the UN flag.
(Note that qualification would not be required if three or more surrogates were used and one
sporadic marginal failure were observed.)
(3) If any surrogate recovery is grossly out of control and the direction of bias is well
defined (i.e., the recoveries of the remaining surrogates are in control or exhibit bias in the same
direction), then qualify the data as follows:
(a) For low bias, qualify all nondetections with the R flag. If an AL is not specified, then
with the X flag.
than the AL with the X flag (e.g., when a conservative estimate is not being sought).
(4) If any surrogate recovery is grossly out of control and the direction of bias is not well
defined, then qualify all nondetections with the R flag. If an AL is not specified, qualify all
detections with the J flag. If an AL is specified, then qualify detections less than the AL with the
X flag. Depending on project DQOs, qualify detections greater than the AL with the J or X flag.
13-5
EM 200-1-10
30 Jun 05
Table 13-1
Data Qualification for Surrogate Recoveries 1
Field Sample
Sample Surrogate Recoveries Result (y) Flag
%R1 and %R2 in control: MRL < MQL < y Flag not required
LCL1 %R1 UCL1
MRL < y < MQL J
LCL2 %R2 UCL2
y < MRL U
%R1 or %R2 marginally OFC with low y > MRL J-
bias:
UN
%R1 < LCL1 or
%R2 < LCL2 y < MRL
%R1 or %R2 marginally OFC with high y > MRL J+
bias:
y < MRL U
%R1 > UCL1 or
%R2 > UCL2
%R1 or %R2 marginally OFC with y > MRL J
inconsistent bias:
%R1 < LCL1, %R2 > UCL2 or
%R1 > UCL1, %R2 < LCL2 y < MRL UN
%R1 or %R2 grossly OFC with low J-
bias: y > MRL X if y < AL
%R1 << LCL1 or
R
%R2 << LCL2 y < MRL
%R1 or %R2 grossly OFC with high y > MRL J+
bias: Possibly, X if y > AL
%R1 >> UCL1 or
y < MRL U
%R2 >> UCL2
%R1 or %R2 grossly OFC with y > MRL J
inconsistent bias: X if y < AL
%R1 << LCL1, %R2 > UCL2 or Possibly X if y > AL
%R1 >> UCL1, %R2 < LCL2 or
%R1 < LCL1, %R2 >> UCL2 or
%R1 > UCL1, %R2 << LCL2 y < MRL R
Notes: 1. It is assumed that the LOI MRL < AL. For the purposes of illustration a field sample result is evaluated
using the recoveries of two surrogates. The subscripts indicate which surrogate is being referenced. For example,
%R1 denotes the percent recovery of the first surrogate. The following abbreviations are used: %R = Recovery of
surrogate spiked into field sample; y = Concentration of a target analyte in the field sample; AL = Action Level;
LCL = Lower control limit for surrogate recovery; UCL = Upper control limit for surrogate recovery; OFC = Out of
control
13-6
EM 200-1-10
30 Jun 05
CHAPTER 14
Data Review Reports
14-1. Introduction.
Post digestion spikes are typically evaluated for trace metal analyses to assess the ability of a
method to successfully recover target metals from an actual sample matrix after the digestion
process has been performed. The PDS results are used with MS results to evaluate matrix inter-
ferences.
14-2. Criteria.
14-2.1. Frequency.
a. Like matrix spikes and matrix duplicates, the frequency of post digestion spikes is
ultimately established from the projects data objectives. No PDSs may be required or a PDS
may be required for every sample in the batch. In general, a PDS should not be required for a set
of environmental samples when a representative MS sample is processed and the MS recovery is
acceptable. Ideally, when a PDS is required, the matrix spike and PDS should be prepared from
the same environmental sample.
Note: Project documents (e.g., QAPPs) often require PDSs to be analyzed at the fre-
quency specified in standard analytical methods (e.g., the CLP SOW requires a PDS for
each sample). Unfortunately, the frequency for PDSs may be poorly or inappropriately
defined some methods. For example, Method 6010B of SW-846 states that a PDS
should be analyzed whenever a new or unusual sample matrix is encountered. How-
ever, the term new or unusual is not well defined. Furthermore, even if new or un-
usual matrix were defined, PDSs would not be required to demonstrate performance if
representative matrix spikes were processed and acceptable MS recoveries were ob-
tained.
b. When project documents do not specify the PDS frequency or the PDS frequency is
deemed to be inappropriate by the reviewer and an unacceptable matrix spike recovery is ob-
served, use professional judgement to determine whether or not a PDS analysis should have been
performed. For example, a PDS would not be required to confirm the presence of matrix inter-
ference if a serial dilution analysis (SDA) were performed and confirms the matrix effect. How-
ever, in the absence of a technically defensible rationale to do otherwise, assume that a post di-
gestion spike (PDS) must be analyzed when the MS is unacceptable.
The acceptance range for each PDS recovery must be no wider than the corresponding accep-
tance range for the matrix spike recovery. When project-specific limits are not specified, an ac-
ceptance range of 85% to 115% is recommended when the concentration of the PDS is at least
14-1
EM 200-1-10
30 Jun 05
two times the native sample concentration. An acceptance range of 80% to 120% is recom-
mended when the spiking concentration is one to two times the native analyte concentration.
Note: The acceptance range for the PDS specified in Method 6010B (75125%) is
wider than acceptance range for the MS (80120%). When metal analyses are per-
formed using Method 6010B and the spiking concentration is high relative to the native
analyte concentration, the acceptance range for the PDS should be no wider than 80
120%.
14-3. Evaluation.
Examine the standard preparation logs to verify that the PDS contains all the target metals. Ex-
amine the sample preparation log to determine whether the PDS was prepared from the same
sample used to prepare the MS. Review the Case Narrative and the PDS summary forms and
note any PDS failures. Using the laboratorys PDS summary form, recalculate the PDS recovery
for at least one target analyte and compare it to the reported value. The reported and calculated
result must agree to within two significant figures.
a. PDS results are qualified using the same strategies for matrix spikes. In particular,
compare the PDS spiking levels to the concentrations of the native analytes in the sample se-
lected for spiking. If the native concentration of a target analyte in the sample (digestate) is
high relative to the spiking concentration, then the PDS recovery may not be representative of
actual method performance. Evaluate the PDS recovery when the spiking concentration is at
least two times greater than the native analyte concentration (e.g., unless the spiking concentra-
tion is slightly less than two times the native analyte concentration and a gross failure occurs). If
environmental samples were qualified (e.g., by the laboratory) for matrix interference but the
spiking level for the PDS is low relative to the native analyte concentration, remove the data
qualifiers.
b. If a single field sample is used to prepare the PDS and MS and the spike
concentrations of both batch QC samples are at least two times greater than the native analyte
concentrations, then evaluate the data as discussed below.
14-4.1. MS Recovery Acceptable and PDS Recovery Unacceptable.
If the MS (and LCS) recovery for a metal falls within the QC acceptance range but a PDS was
analyzed and the PDS recovery is unacceptable, a matrix effect should not be suspected. The
laboratory would normally be expected to reanalyze the PDS sample (digestate) to confirm the
result. Contractual corrective action for unacceptable laboratory performance may be appropri-
ate when a number of failures of this nature are observed and confirmatory reanalyses are not
performed. When a problem of this nature occurs, it is recommended that the Project Manager
be notified. It may be appropriate to request the raw data to perform a more comprehensive re-
view. If there is a gross discrepancy between the PDS and MS recoveries for a particular metal
(e.g., the MS recovery is within 80% - 120% but the PDS recovery is not within 50150%), re-
jection of the data would constitute the most conservative approach.
14-2
EM 200-1-10
30 Jun 05
14-4.2. MS Recovery Unacceptable and PDS Recovery Acceptable.
In general, if the MS recovery for a metal does not fall within the QC acceptance range but the
PDS recovery is acceptable, then a matrix effect (associated with the preparatory process) should
be suspected and the field sample results must be qualified on the basis of the matrix spike re-
covery. However, when historical data for the effect does not exist, the laboratory would nor-
mally be expected to perform a second digestion and reanalysis of the MS to confirm the result.
The result would be confirmed if the MS recoveries and PDS recoveries for both sets of analyses
were similar in magnitude and bias.
14-4.3. MS Recovery Unacceptable and PDS Recovery Unacceptable.
a. When both the MS recovery and PDS recovery for a particular metal fall outside of
QC acceptance range in the same manner (i.e., the PDS and MS failures are of similar magnitude
and the direction of bias is the same), confirmatory analyses are unnecessary. Assume that a
matrix interference exists and use the most noncompliant recovery (the MS or PDS recovery) to
qualify the data.
b. When both the MS and PDS are unacceptable, the laboratory should be expected to
make a reasonable effort to correct for matrix interference before qualifying the field samples for
matrix interference. Review the Case Narrative to determine what corrective actions were per-
formed. Corrective actions for matrix interference may include the use of a different matrix
modifier, different instrument operating conditions, the method of standard additions, internal
standards, a different digestion or analytical procedure, and serial dilutions (if action levels can
be met).
Note: If project documents do not clearly demonstrate that the matrix spike sample is
representative of the samples in the batch (which often occurs in environmental inves-
tigations), then the benefits of extensive corrective actions by the laboratory to mini-
mize a matrix effect should be considered to be minimal. In other words, if the labora-
tory did not make a reasonable attempt to correct for the matrix interference, but the
matrix spike sample is not representative of the samples in the batch, the lack of repre-
sentativeness should be considered to be much more significant than the lack of cor-
rective actions.
c. When the PDS and MS are prepared from two different environmental samples and
the spike concentration is at least two times the native metal concentration, then evaluate the
PDS as follows:
(1) If both the PDS and MS recoveries for a target metal fall outside of the QC
acceptance range in the same manner (i.e., the PDS and MS failures are of similar magnitude and
the direction of bias is the same), then assume matrix interference exists and qualify the data
using the strategies discussed in Chapter 12.
14-3
EM 200-1-10
30 Jun 05
Table 14-1
Evaluation of PDS and MS Data
%R
[MS/PDS] 1 Summary of Evaluation References
PASS / PASS Results not qualified.
1. Check for confirmatory analyses for the digestate
2. Request additional information from the laboratory

PASS / FAIL Chapter 14-4.1
and use professional judgement to either reject the data,
qualify the data using the unacceptable PDS recoveries,
or qualify the data using the acceptable MS recoveries.
1. Check for confirmatory analyses for the matrix
spike.
FAIL / PASS Chapter 14-4.2
2. Qualify for matrix interference based on the MS %R.
1. Check if corrective action was taken to address the
matrix interference.
FAIL / FAIL Chapter 14-4.3
2. Qualify field samples for matrix interference based
upon the most noncompliant of the MS and PDS
recoveries.
Notes:1. It is assumed that the PDS and MS were prepared from the same environmental sample. A PDS or MS
recovery is denoted to be in FAIL status when percent recovery, %R, does not fall within the recovery acceptance
range.
(2) When the PDS recovery is acceptable but the MS recovery is not, use the MS
recovery to qualify the associated field sample results. The laboratory may be required to
analyze additional PDSs for the batch of samples (e.g., one PDS for every sample in the batch)
when the matrix spike recovery is unacceptable. Under these circumstances, verify that the
additional analyses were performed.
(3) When inconsistent PDS and MS recoveries are observed for two different samples in
the preparation batch (e.g., the PDS recovery is biased high and the MS recovery is biased low),
then the representativeness of the PDS and MS results for the remaining samples of batch must
be carefully assessed. In particular, if the MS recovery for a soil sample is acceptable but the
PDS recovery for a second soil sample is not, then the MS and PDS samples may not be repre-
sentative of the remaining samples of the preparation batch and qualification of these samples
may not be appropriate.
14-4
EM 200-1-10
30 Jun 05
CHAPTER 15
Data Review Reports
15-1. Introduction.
a. In general, second column confirmation is required for chromatographic methods

with 2-D detectors when the analytes of concern have not been well characterized. For single-
component (i.e., single-response) analytes, a tentative identification of a target analyte occurs
when the peak associated with the analyte falls within the retention window for the primary
column. A confirmed identification occurs when the analyte peak also falls within the reten-
tion time window for secondary or confirmatory column. The confirmatory and primary
columns must be dissimilar columns (i.e., must possess different stationary phases) so that the
elution order for the target analytes reported from the primary and secondary columns differ.
Target analyte identification for multi component (i.e., multi response) analytes (e.g., Aroclors
by GC) are primary performed using pattern recognition. Hence, second column confirmation
would typically be performed only if the identity of the analyte were in doubt (e.g., would be
performed for weathered Aroclors by GC).
b. A quantitative result from the primary column and a confirmed identification from
the confirmatory column are minimally required for second-column confirmation. In other
words, if the primary column possesses quantitative capability, only detection capability is
minimally required for the confirmatory column. However, it is usually desirable to apply the
same QC criteria to both the primary and confirmatory columns and to report quantitative
results from both analytical columns. (Note, that under these circumstances, the column desig-
nations primary and confirmatory are arbitrary; results reported from either column are
equally reliable.) This strategy is advantageous because it provides a measure of instrument du-
plicate precision. In addition, when a chromatographic interference occurs for the primary col-
umn but does not prevent confirmation, a quantitatively reliable detection may still be reported
from the confirmatory column.
15-2. Criteria.
15-2.1. Frequency.
Unless the analytes of concern have been well characterized or confirmation will be performed
using an instrument with a 3-D detector, second column confirmation must be performed for all
detections (i.e., all results above the reporting limits).
15-2.2. Duplicate Precision.
Unless otherwise specified, assume that quantitative results must be reported from both the pri-
mary and confirmatory columns. If a target analyte is detected with both the primary and con-
firmatory column and the result reported from one (or both) of the columns is greater than the
quantitation limit, then the RPD calculated for the pair of results must be less than or equal to the
absolute value of twice the uncertainty tolerance for the CCVs. In particular, if the error toler-
15-1
EM 200-1-10
31 Jun 05
ance for the CCV is 20% (e.g., 80% - 120%), then the RPD must be within 40% (2 20%).
The RPD for each pair of results is calculated from the equation:
RPD = 100 | y1 y2 | / [( y1 + y2 ) / 2]
where y1 and y2 denote the results from the primary and secondary columns.
15-3. Evaluation.
Verify that all single component analyte detections were confirmed. Confirmation for multi
component analytes will be dependent upon the nature of the contamination and the objectives of
the investigation. All results above the reporting limit must fall within the retention time win-
dows for both analytical columns. If possible, verify that dissimilar chromatographic columns
were used for the primary and confirmatory columns. Calculate the RPD for a pair of results and
ensure that the calculated and reported values agree to within two significant figures.
a. The qualification strategies must distinguish quantitative reliability from qualitative

reliability. If second-column confirmation is required for the project but was not performed,
then, at a minimum, qualify all single-component analyte detections with the N flag (the results
are not qualitatively reliable). Based upon the objectives of the project, the X or R flag may be
more appropriate. If the nature of the site contamination has not been well characterized, then
qualify all the detections with the X or XN flag. For example, if PAH analyses, by liquid chro-
matography with a UV detector, are being performed for a new study area for a risk assessment
and some valid confirmation procedure was not performed, then it would probably be appropri-
ate to qualify detections (especially low-level detections) as tentatively rejected. The X flag
should be used when detections are greater than project-specific action levels and a conservative
estimate is inappropriate (for the particular phase of the project). However, note that, if a sam-
ple is analyzed using second-column confirmation, but an analyte is not confirmed in the sense
that the analyte peak is detected with the primary column but not with the confirmatory column,
then the analyte result is reported as not detected (e.g., using the U flag).
b. At a minimum, qualify all detections with the N flag, if, in the reviewers professional
judgment, the two analytical columns are not sufficiently dissimilar (e.g., a C-18 column is used
with a C-8 column instead of a CN column for explosives by HPLC).
c. Chromatographic interferences from coelutions can affect the quantitative as well as

the qualitative reliability of the data. A high (i.e., noncompliant) RPD may result because one or
more non-analyte peaks that elute in the retention time window for the analyte of interest. Qual-
ify the results with high RPDs as follows:
(1) If the RPD is unacceptable high, at least one of the results is above the method
quantitation limit, and the chromatograms are not available for review, or a coelution cannot oth-
erwise be definitively identified, then, at a minimum, qualify the results from both the primary
and confirmatory columns as qualitatively and quantitatively estimated using the NJ flag. Qual-
15-2
EM 200-1-10
30 Jun 05
ify the results with the X flag when a gross failure occurs and the reason for the unacceptable
RPD is not apparent. However, if a decision level is available and both results are less than the
decision level, then NJ flag may be more appropriate. It is recommended that a gross failure be
defined to occur when the calculated RPD is greater than two times the RPD acceptance limit
(e.g., when the RPD > 80%).
(2) If the result from the primary column and the corresponding result from the
secondary column are both less than the quantitation limit and a high RPD is obtained, then
qualify both results with the J flag (rather than with the N flag).
(3) If it can be determined that a high RPD value arises from a coelution problem but
confirmation is unaffected, then only qualify the result from the column with the coelution
problem as quantitatively estimated or rejected. For example, assume that detections greater
than the quantitation limit for a particular target analyte are reported from both the primary and
confirmatory columns, but the result from the primary column is not quantitatively reliable be-
cause a non-target analyte gives rise to a very large broad shoulder on the target analyte peak.
Since confirmation is unaffected and a quantitative result is available from the confirmatory col-
umn, the result from the confirmatory column would be reported as unqualified (assuming that
all other QC criteria are met), but the result from the primary column may be rejected for quan-
titative reliability. However, it should be noted that a comprehensive data package would typi-
cally be required to perform this type of evaluation.
15-3
EM 200-1-10
31 Jun 05
Table 15-1
Qualification for Second-Column Confirmation 1
RPD Result Reported (Qualified) Result

RPD < 40% MRL < MQL < y1 y1
MRL < MQL < y2
MRL < y1 < MQL
MRL < y2 < MQL y1 J
RPD not calculated because y2 y1 > MRL MRL U
< MRL
MRL < MQL < y1 y1 NJ and y2 NJ 2
40% < RPD < 80% MRL < MQL < y2 or
y2 > y1 y2 NJ
MRL < y1 < MQL y1 J
MRL < y2 < MQL
RPD > 80% MRL < MQL < y1 y1 X and y2 X
MRL < MQL < y2 If y1, y2 < AL, then
y2 > y1 y1 NJ and y2 NJ
MRL < y1 < MQL y1 J
MRL < y2 < MQL
RPD not calculated because MRL < MQL < y1 y1 N or y1 X
confirmation was not
MRL < y1 < MQL y1 JN or y1 X
performed
y1 < MRL MRL U
Notes: 1. Assume both columns are acceptably calibrated and all QC samples are in control (with the possible
exception of the RPD). The result, y1, is being reported from the primary column and result from the confirmatory
column is denoted by y2. The acceptance limit for the RPD is assumed to be 40%. 2. When the RPD > 40 and the
reason for the high RPD is unknown, then the preferred approach is to report the results from both columns. As per
the USEPA OSW memorandum Clarification Regarding Use of SW-846 Methods of 7 August 1998, an
approach that is conservative relative to environmental protection is to report the higher of the two values when the
relative percent difference is greater than 40% and no interferences or chromatographic anomalies are evident.
However, if it can be determined that the high RPD is from a chromatographic interference for one of the columns,
then report the result from the remaining column (unqualified).
15-4
EM 200-1-10
30 Jun 05
CHAPTER 16
Internal Standards for Organic Chromatographic Methods
16-1. Introduction.
For organic analyses, internal standards are compounds that are similar in chemical composition
to the analytes of interest. However, unlike surrogates, internal standards are spiked into all in-
strument QC, batch QC, and environmental samples immediately prior to instrumental analysis.
(Surrogates are spiked into batch QC and environmental samples prior to sample preparation and
analysis.) For environmental applications, the internal calibration technique is typically used for
mass spectrometry methods but may also be used for chromatographic methods with 2-D detec-
tors. Internal standard response should be monitored throughout instrumental analysis to help
evaluate instrument performance (e.g., sensitivity and stability) and matrix effects.
16-2. Criteria.
16-2.1. Frequency.
When quantitation is performed using internal standards, known quantities of internal standards
must be added to all instrument QC, batch QC, and environmental samples immediately prior to
instrumental analysis.
16-2.2. Acceptance Criteria.
a. The compounds used for internal standards, the concentrations, and the acceptance
criteria will be highly dependent upon the analytical technique and the set of analytes of interest.
In general, instrumental response for an internal standard must fall well within the calibration
range. Internal standards in all samples must fall within the retention time windows for the most
recent CCV (especially, for 2-D chromatographic methods). Ideally, internal standards should
also elute at retention times that are near the retention times of the associated target analytes.
b. Unless a more appropriate criterion is available, the peak area for each internal
standard in all instrument QC, batch QC, and environmental samples should be within -50% to
+100% of the corresponding peak area for the mid-level initial calibration standard. The mean
internal standard peak area for the set of initial calibration standards may be used in lieu of the
internal standard peak area of the mid-level initial calibration standard.
Note: This is contrary to the guidance presented in the USEPA Contract Laboratory
Program National Functional Guidelines for Organic Data Review. Internal standard
peak areas for samples are evaluated using the internal standard peak area for the most
recent CCV rather than the internal standard peak areas for the initial calibration. For
example, for the Volatile Data Review, Section X (Internal Standards) states: In-
ternal standard area counts must not vary more than a factor of two (-50% to +100%)
from the associated 12hr calibration standard. This practice is not recommended since
it is based on the assumption that sensitivity, as measured by the internal standard areas
16-1
EM 200-1-10
30 Jun 05
of the CCVs, will not progressively degrade (i.e., decrease) during the analysis. In
other words, it is assumed that the internal standard peak area of each CCV will not
systematically be significantly smaller than that for the preceding CCV.
16-3. Evaluation.
Review the internal standards summary form(s) and ensure that all the internal standards fall
within the appropriate retention time windows and the internal standard areas fall within appro-
priate acceptance limits. Verify, using at least one of the CCVs, that the internal standard peak
area acceptance limits were correctly calculated.
a. In general, environmental samples must be qualified for an unacceptable internal

area count even when the samples are bracketed by acceptable CCVs.
Note: The composition of internal standards and the target analytes are similar but, in
general, it should not be assumed that internal standards will behave in an identical
manner as all the target analytes during all environmental conditions. For example, a
matrix interference for a particular environmental sample may affect the internal stan-
dards more than the associated target analytes (e.g., the preferential absorption of an
internal standard relative to the associated target analytes). Because of differences in
volatility during purge-and-trap analyses, an internal standard may be lost more readily
than its associated target analytes. For example, because of a poor seal for one of the
purge vessels, the internal standard bromochloromethane can be lost more readily than
the associate target analyte acetone.
b. The concentration of a target analyte is inversely proportional to the internal standard

area. Hence, an unacceptably low area count tends to give rise to a high bias and an unaccepta-
bly high area count tends to give rise to a low bias. However, the analyte concentration is ulti-
mately dependent upon the response for the internal standard as well as that for the target ana-
lyte. Therefore, the evaluation of bias on the basis of internal standard response is often
problematic when a comprehensive data package is not available. If the internal standard area
count of a sample does not fall within the acceptance range for internal standard area of the mid-
level initial calibration, then qualify the results as summarized in the table below. It is assumed
that a high internal standard response gives rise to a low bias. However, in order to obtain a
more conservative evaluation of the data, a direction of bias is not assumed for low internal stan-
dard response.
c. An internal standard peak area that does not meet the threshold criteria for a detection
may give rise to a false negative for the associated target analyte and surrogate results. If an ex-
tremely low area count is reported and the chromatograms are not available for review (e.g., to
assess signal to noise ratios), a conservative approach must be used; qualify nondetections with
the R flag. However, if a more detailed review of the data is planned, then the X flag would be
more appropriate.
16-2
EM 200-1-10
30 Jun 05
d. A more detailed review of the data is recommended when a samples internal standard
does not fall within the retention time acceptance windows. For example, the data evaluator
should request a comprehensive data package and the raw data (e.g., chromatograms and quan-
titation reports) should be examined to determine if any false positives or false negatives exist.
Table 16-1
Data Qualification of Internal Standard Areas 1
Internal Standard Area of Sample

(A) Sample Result (y) Flag
MRL < MQL < y None
MRL < y < MQL J
A0 A 2 A0
y < MRL U
y > MRL J-
2 A0 < A < 5 A0
y < MRL UN
y > MRL J
(A0 /5) < A < A0
y < MRL U
J-
A > 5 A0 y > MRL X if y < AL
y < MRL R
y > MRL X
A < (A0 /5)
y < MRL R
Notes: 1.I t is assumed that the LOI MRL < AL. The following abbreviations are used: A = Internal standard area
count sample; A0 = Internal standard area count sample; y = Concentration of a target analyte for the field sample;
AL = Action Level; MRL = Method Reporting Limit; MQL = Method Quantitation Limit.
16-3
EM 200-1-10
30 Jun 05
APPENDIX A
Glossary
A-1. Accuracy.
Accuracy refers to closeness to the true value. The true value typically refers to the expected or
prepared spike concentration. For statistical applications, the true value typically refers to the
population mean. Accuracy is a conceptualization and cannot be known with complete certainty.
Accuracy is evaluated in a variety of ways. For laboratory analyses, accuracy is commonly
inferred from the percent recoveries of spike samples (matrix spikes, laboratory control samples,
surrogate spikes, performance evaluation samples, etc.).
The terms accuracy and bias are commonly used interchangeably, but, strictly speaking,
these terms are not synonymous. Accuracy is generally a function of both random error and
systematic error. Random error is characterized by unpredictable variations for the
measured results of a parameter of interest, cannot be corrected directly, but can be reduced by
increasing the number of measurements. (It should be noted that random error is distinguished
from spurious error, which is also unpredictable, but arises from factors such as human
blunder and gross instrument malfunction.) The term bias refers to systematic directional error
from the true value. Unlike random error, bias or systematic error remains constant or
varies in a predictable manner, but is independent of the number of measurements. The mean
spike recovery for the LCS is a measure of method bias and the standard deviation for the LCS
recoveries is a measure of method precision. The mean LCS recovery with the standard
deviation essentially constitutes a measure of accuracy.
A-2. Analyte.
See Target Analyte.
A-3. Batch.
See Preparation Batch.
A-4. Batch Quality Control (QC) Sample.
See Quality Control (QC) Sample.
A-5. Bias.
Bias refers systematic error. Bias is a directional error that arises from a constant or
predictable distortion of the measurement process. A measurement or estimate is said to be
unbiased if the mean approaches the true value as the number of replicates increases. An
estimate is said to possess a low bias if it is consistently less than the true value and is said to
possess a high bias if it is consistently higher than the true value. The adjectives high and
low are used to refer to the direction rather than the magnitude of the deviation from the true
A-1
EM 200-1-10
30 Jun 05
value. Adjectives such as small, slight, marginal, large, and gross will be used to
refer to the magnitude of the deviation. For chemical analyses, consistently low or high
recoveries for batch QC samples (e.g., laboratory control samples and matrix spikes) are
indicative of bias.
A-6. Chain-Of-Custody (COC).
Chain-Of-Custody (COC) procedures and forms primarily document the possession of the
samples from collection to storage, analysis, reporting, and, ultimately, disposal. Each cooler
sent from the field to a laboratory is accompanied by a unique COC record. (The COC form is
typically sealed in a Ziploc-type bag and is taped to the inside of the cooler lid.) COC forms
must become part of the permanent record of all sample handling and shipment. The COC form
lists the samples in a cooler, and includes the following information: project identification,
unique project-specific sample identifications, dates and times of sample collection, number of
containers, general testing procedures, and any special remarks. Couriers shipping documents
should also be included.
A-7. Characteristic Peaks.
For multi component target analytes (e.g., Aroclors), characteristic peaks are those peaks that
are at least 25% of the height of the largest peak in chromatogram for the pure multi-component
standard.
A-8. Comparability.
Comparability refers to the equivalency of two sets of data. This goal is achieved through the
use of standard or similar techniques to collect and analyze representative samples. Comparable
data sets must contain the same variables of interest and must possess values that can be
converted to a common unit of measurement. Comparability is normally a qualitative parameter
that is dependent upon the other data quality elements. For example, if the detections limits for a
target analyte were significantly different for two different methods, the two methods would not
be comparable.
A-9. Completeness.
Completeness refers to the percentage of data that is valid or usable; that is, which satisfies
project-specific DQOs. The highest degree of completeness that can be achieved is normally
desired. Completeness acceptance criteria would normally be defined for both field and
laboratory activities. A typical acceptance criterion for completeness is 80% to 90%. A higher
completeness acceptance criterion may be required for critical samples. In general, when
calculating percent completeness, R-qualified and X-qualified data must not be included in the
set of valid data.
A-10. Comprehensive Blank.
See Hierarchy of Blanks.
A-2
EM 200-1-10
30 Jun 05
A-11. Comprehensive Data Package.
A comprehensive data package is defined as a package of definitive or effective chemical

data that satisfies the minimum data reporting requirements of this document and contains
sufficient information to completely reconstruct the chemical analyses that were performed.
Comprehensive data packages include all batch, method, and instrument QC results as well as
raw data (e.g., run logs, sample preparation logs, standard preparation logs, and printed
instrumental output such as chromatograms).
A-12. Continuing Calibration Verification (CCV).
A continuing calibration verification (CCV) refers to the use of a mid- to low-level instrumental
standard to check rather than to alter instrument calibration. CCVs are typically analyzed on a
continuing basis (e.g., at the beginning, middle, and end of an analytical sequence) and are
evaluated to determine whether the instrument was within acceptable calibration throughout the
period of time samples were instrumentally analyzed. The CCV is usually (but necessarily)
prepared from a standard that is from the same source as the initial calibration standards.
A-13. 2-D and 3-D Detectors.
A 3-D detector differs from a 2-D detector in that the former furnishes quantitative and
comprehensive qualitative information for definitive compound identification, while the latter
primarily furnishes only quantitative information. Detectors such as PIDs, ECDs, and FIDs are
referred to as 2-D or two-dimensional detectors since they essentially yield a two-
dimensional plot of gross instrumental response versus time (i.e., single-channel time-versus-
response data). Two-dimensional detectors cannot provide sufficient qualitative information for
analyte identification. Detectors such as mass selective and infrared (IR) detectors are examples
of 3-D or three dimensional detectors since they provide time-versus-response data for
multiple mass ions and wavelengths, respectively.
A-14. Data Quality Indicators.
See PARCCS.
A-15. Data Quality Objectives (DQOs).
Data quality objectives (DQOs) are perhaps more appropriately referred to as data objectives.
DQOs refer to the quantitative and qualitative statements that identify the goals, decision
strategies, and boundaries for a particular study (e.g., acceptable levels of uncertainty); in
addition, DQOs define the type, quality, and quantity of data required to support project
decisions by the data users. The DQOs are developed during the planning stages of a project
based upon the scientific method of inquiry. With respect to the chemical testing, DQOs are
developed prior to sample collection and analysis, in order to determine appropriate analytical
methodology, quality control acceptance limits (i.e., specifications for data quality indicators),
and corrective actions.
A-3
EM 200-1-10
30 Jun 05
A-16. Detection.
A detection typically refers to a reported measured concentration of a target analyte that is

greater than the detection limit or the reporting limit.
A-17. Definitive Data.
The distinction between definitive and screening data is rather subjective. Definitive data are
typically produced using rigorous analytical methods, such as EPA reference methods. The
analytical results are frequently evaluated with respect to relatively stringent quality control
specifications and PARCCS criteria are well defined. Recently, the term effective data (versus
definitive data) has been used to describe data of sufficient quality to support project
decisions. Screening data are essentially data that are not fully effective--data that cannot be
used to support project decisions without higher quality data.
For example, since screening methods often lack specificity, they tend to give rise to false
positives. Therefore, screening data are usually confirmed by testing a percentage of the
environmental samples (e.g., 10%) with definitive methods or more effective methods of
analysis. Quantitative data from screening methods also tend to be less precise and accurate than
that from definitive methods. Screening data are typically generated by methods of analysis that
are relatively rapid (typically involving minimal sample preparation) and performed in the field
(as opposed to an off-site laboratory). However, real-time data generated in the field is not
necessarily of inferior quality to fixed laboratory data.
A-18. External Calibration.
The external calibration technique is primary used for organic chromatographic analyses
involving detectors other than MS detectors (e.g., FID, PID, ECLD, ECD and NPD). A
calibration factor is calculated for each analyte and surrogate in each initial calibration standard
using the equation:
CF = Peak Area or Height of analyte in calibration standard

Amount of the Compound Injected (e.g., mass in nanograms)
For multi component analytes, the numerator is the sum of the area or heights of several peaks.
In other words, the calibration factor is the ratio of detector response to the amount of analyte in
the calibration standard. The amount of analyte in an environmental sample is calculated by
dividing the instrumental response for the analyte by the mean calibration factor for all the initial
calibration standards.
A-19. Field Duplicates.
Field Duplicates are similar to matrix duplicates. They differ in that the former are prepared in
the field while the latter are prepared in the laboratory. A field duplicate is an environmental
sample that is homogenized and split into two separate aliquots in the field rather than at the
A-4
EM 200-1-10
30 Jun 05
laboratory. This document distinguishes between field duplicates and collocated samples.
Collocated samples are field samples that are collected near each another during a single
sampling event but which are not homogenized. However, for simplicity, field duplicates and
colocated samples are not distinguished from laboratory duplicates when homogenization cannot
be performed because of the nature of the analyte or the methodology. For example, samples
collected for low-level VOC analysis by closed-system purge-and-trap cannot be homogenized.
Hence, for these type of analyses, the term matrix duplicate refers to collocated samples.
A-20. Field QC Samples.
Field QC samples are QC samples that are prepared in the field or that are impacted by field
activities. Examples of field QC samples include trip blanks, rinsate (equipment) blanks, and
field duplicates. Matrix spike samples may or may not be field QC samples. For example, if an
environmental sample were homogenized in the field and subsequently split into three aliquots
for MS and MSD analyses, then the MS and MSD samples would be considered to be field QC
samples.
A-21. Hierarchy of Blanks.
When environmental (field) samples are subjected to multiple handling, preparatory and
analytical procedures, blanks may be introduced in a sequential manner to measure the level of
contamination arising from each procedure or from select sets of procedures. For example,
assume that a sample is sequentially processed using two distinct preparatory techniques, which
will be referred to as technique 1" and technique 2. The sample and a blank, BLK(1,2), are
processed using technique 1. The sample, the blank BLK(1,2), and a second blank, BLK(2), are
then processed using technique 2. When blanks are processed in this manner, they can be used
to evaluate the contamination associated with each stage of the preparatory process. For
example, assume that contaminants are detected in BLK(1,2), but none are detected in BLK(2).
It would be assumed that the contamination resulted from the first preparatory technique.
Furthermore, since BLK(1,2) accounts for the contamination introduced from the entire
preparatory process, only BLK(1,2) would minimally be required to evaluate environmental
samples for contamination. For example, if BLK(2) were not processed, then the samples would
be evaluated using BLK(1,2) alone.
A blank that measures contamination for a set of handling, preparatory, or analytical procedures
is said to possess a higher hierarchy than a blank that measures contamination for only a subset
of the procedures. In the example cited above, the blank BLK(1,2) possesses the highest
hierarchy because it measures contamination from all of the preparatory techniques. The highest
hierarchy blank will be referred to as a comprehensive blank if it accounts for contamination
from all sample handling, preparatory, and analytical procedures. In general, a blank with a
higher hierarchy is more critical than one with a lower hierarchy. In particular, when a blank is
missing for a set of environmental samples, the samples may be qualified for contamination
using a blank that possesses a higher hierarchy than the missing blank. Blanks for environmental
analyses are listed in order of increasing hierarchy below:
Calibration/Instrument Blanks < Storage/Holding Blanks < Method Blanks
A-5
EM 200-1-10
30 Jun 05
< Trip Blanks < Rinsate/Equipment Blanks
Note that trip blanks are collected only for VOC analyses. Furthermore, a rinsate blank may be
substituted for a trip blank only when the rinsate blank is stored and shipped in the same cooler
as the field samples. Under these circumstances, the rinsate blank would account for
contamination arising from cleaning procedures, cross contamination in the sample coolers, and
laboratory contamination.
A-22. Holding Time.
The preparation holding time (e.g., the extraction or digestion holding time) is defined as the
period of time from the date an environmental sample is collected in the field to the date the
sample is processed with the preparatory method (e.g., the date the sample is first exposed to the
extraction or digestion solvent). The analysis holding time is defined as the period of time from
the date of sample preparation (e.g., extraction or digestion) to the date of sample analysis using
some determinative (i.e., instrumental) method.
A-23. Holding Time Limit.
The holding time limit is defined as the maximum acceptable holding time for sample
preparation or analysis.
A-24. Initial Calibration.
Initial calibration refers to the establishment of a quantitative relationship between instrumental

response and analyte concentration (or amount) prior to the analysis of samples. The correlation
between instrumental response and analyte concentration is established via the analysis of a set
of standards of known concentration and is demonstrated using quantitative performance
specifications (e.g., linear correlation coefficients). The initial calibration must demonstrate that,
over some concentration range of interest, a change in analyte concentration is associated with a
predictable change in instrumental response and vice versa (i.e., there is a continuous functional
and inverse functional relationship between instrumental response and concentration).
A-25. Initial Calibration Verification (ICV).
An initial calibration verification (ICV) refers to the use of a mid-level, second-source,

instrumental standard to verify the accuracy of the standards used to perform the initial
calibration. The ICV is typically performed immediately after the initial calibration. The
acceptance limits for ICV recoveries should be similar to the acceptance limits for other
instrumental QC samples such as CCVs.
A-26. Instrument Quality Control (QC) Sample.
See Quality Control (QC) Sample.
A-6
EM 200-1-10
30 Jun 05
A-27. Internal Calibration.
The internal calibration technique is primarily used for organic chromatographic analyses
involving MS detectors. An internal standard is added to each sample and calibration standard
immediately prior to analyses. An internal standard is a substance that is similar to the target
analytes in chemical behavior, which is not normally found in the environmental samples, and
which is added at a fixed, known concentration to all samples and calibration standards. A
relative response factor is calculated for each analyte and surrogate in each initial calibration
standard according to the equation:
RRF = As Cis
Ais Cs
where
As = Peak area or height of the analyte or surrogate

Ais = Peak area or height of the internal standard
Cs = Concentration of the analyte or surrogate (e.g., g/L)
Cis = Concentration of the internal standard (e.g., g/L)
For multi component analytes such as dioxins, the terms As and Ais represent the sum of the
integrated ion abundance of multiple quantitation ions. The calculation of the amount of analyte
in an environmental sample involves dividing instrumental response for the analyte by the
instrumental response for the internal standard and the mean relative response factor for the set
of initial calibration standards. The internal standard technique is superior to the external
standard technique because target analyte loss is taken into account for the portion of the
analytical process that takes place after the internal standard is spiked into the sample (e.g., loss
during sample injection). The internal standard procedure is used primarily with MS detectors
because the signal intensities used for quantitation would not otherwise be adequately stable and
the masses of the internal standards can be resolved from those of the target compounds even
when chromatographic resolution cannot be achieved.
A-28. Laboratory Control Sample (LCS).
The laboratory control sample (LCS) is used to assess laboratory method performance.
Laboratory control samples are analyzed to assess the ability of the laboratory to successfully
recover the methods target analytes from a sample of known composition. Precision may be
evaluated by analyzing an LCS and an LCSD (laboratory control sample duplicate) for each
preparation batch of samples. A laboratory control sample must be chemically and physically
similar to the environmental samples and must contain a known amount of each target analyte at
an appropriate concentration. A laboratory control sample typically consists of a clean matrix
(e.g., reagent water or purified sand) that has been spiked with the target analytes of interest. In
general, an LCS must contain all single-component target analytes of interest and must be
processed through the entire sample preparatory and analytical methods. The LCS usually
contains only a subset of the target analytes when multi component analytes such as Aroclors are
A-7
EM 200-1-10
30 Jun 05
being analyzed. Ideally, the concentrations of the target analytes in the LCS should be
determined by project-specific DQOs (e.g., should be near the regulatory or risk-based decision
limits), but the LCS is typically spiked between the low-level and mid-level calibration
standards.
Internal LCS acceptance limits for accuracy and precision are produced by the laboratory by
performing statistical calculations (using at least 30 data points). However, since duplicate
precision is not as critical as accuracy, many laboratories do not generate statistical acceptance
limits for duplicate precision but use default (e.g., method specified or arbitrary) RPD
acceptance limits. Control charts and tables are maintained to establish the bias and precision of
the method, and are updated periodically (typically, on a quarterly basis). A representative
subset of the target analytes for each method is normally graphed to observe method trends.
Unfortunately, LCS acceptance limits for environmental sampling and analysis activities are
often based upon the laboratorys internally generated control chart limits or method-specified
limits rather than project-specific DQOs. Ideally, project-specific acceptance limits should be
equal to or greater than the laboratorys in-house statistical control limits.
When an LCS result falls outside of the laboratorys internal acceptance limits, the laboratory
must implement some form of corrective action. In general, the preparation batch must be
reprocessed when the associated LCS recovery falls outside of the acceptance range. When an
LCS RPD is out-of-control but the LCS recovery is acceptable, the laboratory must implement
corrective action but the associated environmental samples would not typically be reprocessed.
A-29. Laboratory Control Sample Duplicate (LCSD).
See Laboratory Control Sample (LCS).
A-30. Limit of Identification (LOI).
The limit of identification (LOI) is the lowest concentration of analyte that can be detected with
99% confidence; that is, the LOI is the concentration at which the probability of a false negative
(Type II decision error) is 1%. The LOI is adjusted for method specific factors (e.g., sample
size) and may be approximated as twice the detection limit. The LOI may be set equal to about
two times the MDL (e.g., if it is assumed that the standard deviation is not strongly dependent
upon concentration).
A-31. Matrix-Dependent Duplicate.
See Matrix Spike Duplicate (MSD) and Matrix Duplicate (MD).
A-32. Matrix Spike (MS).
The matrix spike (MS) is used to assess the performance of the method as applied to a particular
matrix; they are analyzed to assess the ability of the method to successfully recover target
analytes in the environmental population being sampled. An MS is an environmental sample to
which known concentrations of all of the method target analytes have been added before it is
A-8
EM 200-1-10
30 Jun 05
carried through all sample preparation, cleanup, and analytical procedures. MS results are
evaluated in conjunction with other QC information (e.g., surrogate and LCS recoveries) to
determine the effect of the matrix on the accuracy of the analysis.
The target analytes added to a matrix spike sample would typically be identical to those added to
a laboratory control sample. The environmental sample selected for MS analysis must be
representative of the environmental population being sampled and would normally be specified
in the field. Control charts may be maintained for MS recoveries, but, in general, laboratories do
not base batch control on the results of MS samples unless a general method failure is indicated.
Matrix spikes are typically analyzed at a frequency of at least 5% but frequency requirements are
project-specific.
A-33. Matrix Duplicate (MD) and Matrix Spike Duplicate (MSD).
The matrix duplicate (MD) and matrix spike duplicate (MSD) are used to assess the precision of
a method in an actual matrix. A MSD is a duplicate of an MS. An MSD is also used to evaluate
the accuracy of a method in a particular matrix and is evaluated using the same criteria for the
MS. An MD is an environmental sample that is divided into two separate aliquots. (Care must
be taken to ensure that the sample is properly divided into homogeneous fractions.) Both the
MD and MSD are carried through the complete sample preparation, cleanup, and analytical
procedures. For brevity, MS/MSD and matrix duplicate pairs are referred to as matrix-
dependent duplicates.
Frequency requirements for MDs and MSDs are normally established on a project-specific basis.
An MD is normally processed with each preparation batch when target analytes are expected to
be present. A MSD is normally processed with each preparation batch when method target
analytes are not expected to be present. As a rule of thumb, a MSD is used for organic
methods and a MD is used for inorganic methods. The results of the MD or MSD are evaluated,
in conjunction with other QC information, to determine the effect of the matrix on the precision
of the analysis. Control charts, or tables, may be maintained for these samples to monitor the
precision of the method for each particular matrix and may be required by certain projects.
A-34. Matrix Interference.
As used in this document, the term matrix interference typically refers to an effect that arises
from the native physical or chemical composition of an environmental sample that produces a
negative or positive bias in the results.
For example, high concentrations of non-target analytes that coelute with the analytes of interest
in the instrumental portion of a chromatographic method may give rise to a positive interference
(i.e., high bias). Substances such as peat and clay may bind the target of interest and prevent
complete extraction of the target analytes in the preparatory portion of an analytical procedure
(especially when analyte concentrations are low), may give rise to a negative interference (i.e.,
low bias). However, sample heterogeneity is viewed as a characteristic of the matrix (e.g., the
spatial variability of the environmental population being sampled) rather than as an
interference for which the method of analysis must be optimized to reduce.
A-9
EM 200-1-10
30 Jun 05
A-35. Measurement Quality Objectives (MQOs).
Measurement quality objectives are acceptance criteria for PARCCS for the various phases of
the measurement process (e.g., sampling and analysis) that are established to ensure that total
measurement uncertanity is within the range prescribed by project DQOs.
A-36. Method Blank (MB).
Method blanks are used to assess laboratory contamination. A method blank is defined as an
interference-free matrix which is similar to the field sample matrix, lacks the target analytes of
interest, and is processed with the environmental samples using the same preparatory and
determinative methods. Hence, all reagents added to samples during extraction, cleanup, and
analysis are also added to method blanks in the same volumes or proportions. Analyte-free
reagent water is frequently used to prepare method blanks for aqueous analyses and a purified
solid matrix (e.g., sand) is frequently used for solids.
Contamination may result in false positives or elevated reporting levels for target analytes.
Method blanks are analyzed to assess contamination for the entire analytical process. Therefore,
when a batch of samples is analyzed on separate instruments or separate analytical shifts, the
method blank associated with the batch (e.g., extracted with the samples) must also be analyzed
with the samples for each instrument and analytical shift.
A-37. Method Detection Limit (MDL).
The method detection limit (MDL) is the minimum concentration of a substance that is
significantly greater than zero (an analytical blank) at the 99% limit of confidence and is
determined using the procedure described in 40 CFR, Part 136, Appendix B. The standard
deviation, s, is calculated for n replicate aliquots (where n > 7) that are spiked near (e.g., one to
five times) the estimated MDL and processed (as environmental samples) through the full
analytical procedure. The standard deviation for the set of replicate analyses is subsequently
multiplied by the Student t value corresponding to the 99% percentile of the t-distribution with
n-1 degrees of freedom.
Since it is not practical to establish an MDL for each specific matrix received at any given
laboratory, MDLs are usually estimated in interference-free matrices (typically reagent water for
aqueous analyses and a purified solid matrix such as sand for the analysis of solid matrices).
However, certain projects may require the determination of method detection limits in site-
specific matrices.
As defined in 40 CFR 136 Appendix B, MDLs are method, matrix and instrument specific.
MDL samples must be processed using the sample determinative and preparatory methods as the
environmental samples (e.g., using the same extraction and cleanup procedures) and must be
adjusted for method-specific procedures such as dilutions. When multiple instruments are used
to perform the same method, MDLs may be demonstrated on individual instruments (including
individual chromatographic columns and detectors) via the analysis of MDL check samples.
A-10
EM 200-1-10
30 Jun 05
An MDL check sample is prepared by spiking an interference free matrix with all target analytes
of interest at about two times the estimated MDL and subsequently processing it through the
entire analytical procedure. If a target analyte is not recovered in the MDL check sample, then
the MDL study should be repeated for that target analyte. It is recommended that a laboratorys
MDLs be verified quarterly by analyzing detection limit check samples. MDL studies should be
performed at least annually and whenever the basic chemistry or instrumentation for method is
changed.
It should be noted that the statistical approach described in 40 CFR 136 Appendix B does not
take calibration uncertainty into account. It is implicitly assumed that the calibration curve is
known with certainty. Since the variability associated with the estimate of the calibration curve
is not taken into account, when samples to not undergo a significant preparatory process, it may
be desirable to establish detection limits using the procedure described by Andre Hubaux and
Gilbert Vos (Decision and Detection Limits for Linear Calibration Curves, Analytical
Chemistry, Volume 42, No. 8, July 1970).
A-38. Method Quantitation Limits (MQL).
The method quantitation Limit (MQL) is the concentration of an analyte in a sample that is
equivalent to the concentration of the lowest initial calibration standard adjusted for method-
specified sample weights and volumes (e.g., extraction volumes and dilutions). Typically,
MQLs are equal to or greater than the lowest initial calibration standard and are at least five
times greater than the MDL. MQLs must also be less than project-specific action levels. It is
usually desirable for the MQL to be equal to some fraction of the projects action levels (e.g.,
one half or one third of the action levels).
A-39. Method Reporting Limit (MRL).
The method reporting limit (MRL) is the threshold or censoring limit below which target analyte
concentrations are reported as < MRL or MRL U, where MRL is the numerical value of
the method reporting limit. The method reporting limit is usually established based on the
laboratory's LOIs, MQLs, or project-specific action levels. The MRL for undetected analytes
should not be less than the LOI or RDL and must not be greater than the AL.
A-40. Native Analyte.
In the context of environmental testing, the term native analyte refers to the analyte
incorporated into the test material by natural processes or from past waste handling activities
(e.g., as opposed to spike addition).
A-41. Nondetection.
A nondetection typically refers to a target analyte concentration that is less than the detection
limit or the method reporting limit.
A-11
EM 200-1-10
30 Jun 05
A-42. PARCCS.
The term PARRCS is an acronym for the primary elements of data quality: Precision,
accuracy, representativeness, completeness, comparability, and sensitivity. It should be noted
that sensitivity is often omitted and the acronym PARCC is commonly used; these five data
quality elements (PARCC) are often referred to as Data Quality Indicators (DQIs).
A-43. Percent Difference (%D).
The percent difference of a measurement, Xo, of a variable X is defined by the equation:
%D(X) = ( | Xo X | / X ) 100
where
X = Xi / n
is the mean of a set of n replicate measurements of X (that excludes Xo). For brevity, unless
otherwise specified, the term percent difference (%D) will refer to the percent difference for the
response factor of a continuing calibration standard for an organic chromatographic method.
A-44. Percent Recovery (%R).
The percent recovery for a matrix or post-digestion spike is defined by the equation:
%R = 100 (XF XO) / S
where
XF = Measured concentration of environmental sample after spike addition

XO = Measure concentration of environmental sample prior to spike addition
S = Spike (reference) concentration
For CCVs, ICVs, and LCSs, the percent recovery is defined as:
%R = 100 (X / S)
where
X = Measured concentration of QC sample
For brevity, the percent recovery is referred to as the recovery.
A-12
EM 200-1-10
30 Jun 05
A-45. Percent Relative Standard Deviation (%RSD).
The percent relative standard deviation for n replicate measurements of a variable X is defined
by the equation:
%RSD(X) = (SD /X ) 100
where
X = Xi / n
is the mean of the variable. For brevity, the term relative standard deviation (%RSD) will refer
to the percent relative standard deviation of the response factors for the initial calibration
standards for a chromatographic method.
A-46. Performance-Based Method/Approach.
This term does not appear to be well-defined in the literature. As applied to chemical testing, the
term performance-based implies that the methodology used to produce an analytical result is
secondary to the quality of the result itself. When a performance-based approach is
implemented, specifications are primarily imposed upon the data (the end product of the
analytical process) rather than upon the process by which the data are produced. Chemical data
are generated by any analytical method which can demonstrate project-specific PARCCS
requirements are met. Method QC elements such as detection limits, method blanks, laboratory
control samples, and matrix spikes are minimally required to demonstrate method performance.
A-47. Post Digestion Spike (PDS).
A post digestion spike (PDS) is typically analyzed for metals to assess the ability of a method to
successfully recover target metals from an actual matrix after the digestion process. A PDS is an
environmental sample to which known concentrations of target metals are added after the
digestion process. The spiking concentration for the PDS should not be less than about two
times the native analyte concentration. The same target analytes should be spiked into the LCS,
MS, and PDS. A PDS should be analyzed when the MS is unacceptable. When the MS is
unacceptable, an aliquot of the same environmental sample should be selected for the PDS.
Alternatively, a PDS should be routinely processed with each MS so that every batch of samples
contains at least one sample that is spiked before and after the digestion process.
A-48. Precision.
Precision refers to the repeatability of measurements. For statistical applications, precision

refers to the spread or distribution of values about the population mean and is frequently
measured by the standard deviation. For the chemical analyses of environmental samples,
precision is commonly determined from duplicate samples (e.g., matrix spike duplicates, matrix
A-13
EM 200-1-10
30 Jun 05
duplicate and laboratory control sample duplicates) and is commonly measured using either the
relative percent difference (RPD) or the percent relative standard deviation (%RSD).
A-49. Preparation Batch.
A preparation batch is defined as a set of samples that are prepared together by the same person
or group of people; using the same equipment, glassware, and lots of reagents; by performing
manipulations common to each sample in the same sequence and within the same time period
(usually not to exceed one analytical shift). Ideally, the samples in a preparation batch must be
from the same study area and must be of similar composition. Samples taken from the same
study area would normally be grouped together for batching purposes within the constraints
imposed by the method holding times. However, laboratories may find it necessary to group
samples from different clients into a single batch.
Environmental and QC samples must be prepared, analyzed, and reported in a manner that is
traceable to individual batches. Hence, each preparation batch must be uniquely identified
within the laboratory. A preparation batch is normally limited to twenty field environmental
samples of a similar matrix and also contains the appropriate QC samples (e.g., a laboratory
control sample and a method blank). The QC samples undergo the same preparatory procedures
(e.g., using the same extraction and cleanup methods) as the environmental samples. Samples in
the same preparation batch would normally be analyzed together using the same instrument.
A-50. Preservation.
The term preservation refers to any technique (frequently involving the addition of laboratory-
grade reagents) that retards biological, chemical, or physical processes that would alter the
representativeness of the sample relative to the environmental population of interest (e.g., alter
the analyte concentration in the sample matrix being tested). The most common preservation
methods include pH adjustment, dechlorination, and temperature adjustment (i.e., cooling or
freezing).
A-51. Professional Judgment.
As per ISO/IEC Guide 25 (August 1996 draft), the term professional judgment refers to the
ability of a single person or a team to draw conclusions, give opinions and make interpretations
based on measurement results, knowledge, experience, literature and other sources of
information. A professional judgment must be supported by appropriate documentation. The
information or factors taken into account during the decision making process must be discussed.
A-52. Quality Control (QC) Sample.
This document distinguishes between preparatory methods (e.g., Method 3010A) and
determinative methods (e.g., Method 6010A) of analyses. A QC (quality control) sample that is
independent of matrix effects and analyzed only in a determinative method is referred to as an
instrument QC sample (e.g., a CCV and CCB). A non-instrument QC sample that is processed
with the same preparatory and determinative methods as the environmental samples (e.g., matrix
A-14
EM 200-1-10
30 Jun 05
spikes and laboratory control samples) is referred to as a method QC sample. Note that method
QC samples (e.g., MDL study samples) are not necessarily analyzed on a per batch basis. A
non-instrument QC sample that is analyzed on a per batch basis is referred to as a batch QC
sample. Hence, a batch QC sample is a method QC sample that is analyzed on a per batch basis,
or is a QC sample which is analyzed in only the determinative method but which is dependent
upon matrix effects (e.g., post-digestion spikes).
A-53. Recovery.
See Percent Recovery.
A-54. Relative Percent Difference.
The relative percent difference for a set of duplicate measurements of the variable X, RPD(X), is
defined by the equation:
RPD(X) = ( |X1 X2| / X ) 100
where
X = (X1 + X2) / 2
is the mean of the pair of variables. The RPD is a measure of precision. For brevity, unless
otherwise specified, the term relative percent difference refers to the relative percent difference
of duplicate spike recoveries.
A-55. Reliable Detection Limit (RDL).
The reliable detection limit (RDL) is the upper 95% upper confidence limit of the MDL defined
in 40 CFR, Part 136, Appendix B. RDLs for the upper 100% upper confidence limit are
calculated as follows:
RDL = [ (n 1) / n 1, (1 ) / 2 ] MDL
where the MDL is calculated by multiplying the standard deviation, s, by the 99th percentile
point of the t-distribution with n1 degrees of freedom, tn1, 0.99:
MDL = tn 1, 0.99 s
The number of replicate analyses used to compute the MDL is denoted by n. The 100 (1 ) / 2
percentile of the Chi-Square distribution with n - 1 degrees of freedom is denoted by
n - 1, (1 ) / 2. For n = 7 and = 0.95
n 1, (1 ) / 2 = 6, 0.025 = 1.24
and
A-15
EM 200-1-10
30 Jun 05
RDL = (6 / 1.24) MDL 7 s 2 MDL
A-56. Representativeness.
Representativeness refers to the degree to which a sample or set of samples estimates the char-
acteristics of a target population. For the chemical analysis of environmental samples, repre-
sentativeness is a usually a qualitative parameter that is dependent upon the design of the field
sampling program and laboratory methods (e.g., subsampling techniques). An evaluation of rep-
resentativeness would include an assessment of laboratory holding time and method blank data.
For example, samples that are not properly preserved or that are analyzed beyond acceptable
holding times may not provide representative data.
A-57. Response Factor.
The term response factor refers to the calibration factor or relative response factor. Refer to
internal calibration and external calibration.
A-58. Rinsate Blank.
Equipment or rinsate blanks consist of reagent water passed through or over sampling equipment
following sample collection and sample equipment decontamination. Contaminated equipment
blanks indicate inadequate decontamination between samples and a likelihood of cross-
contamination between samples.
A-59. Sample.
The term sample refers to non-instrument QC samples (i.e., batch QC and method QC
samples) and environmental (field) samples.
A-60. Sensitivity.
Sensitivity refers to the amount of analyte necessary to produce a detector response that can be
reliably detected or quantified. Detection limits (e.g., instrument and method) and quantitation
limits are commonly used to measure sensitivity.
A-61. Surrogate.
In the context of environmental testing, a surrogate is a relatively pure organic compound which
is added to samples prior to preparation and analysis and which is similar to the analytes of
interest (in physical and chemical behavior), but which is not normally found in environmental
samples. Surrogates are typically spiked into environmental samples as well as batch QC and
instrument QC samples for chromatographic methods. Surrogate recoveries in environmental
samples are primarily used to assess overall performance on a sample-specific basis. Surrogate
recoveries for environmental samples measure matrix effects (e.g., and extraction efficiency for
A-16
EM 200-1-10
30 Jun 05
organic analysis involving solvent extractions), are evaluated in a similar manner as matrix
spikes, but are evaluated on a sample-specific rather than batch-specific basis.
Surrogate recoveries for instrument QC samples (such as continuing calibration standards) are
dependent upon instrument performance. Surrogate recoveries for the LCS and MB are used to
evaluate the performance of the preparatory and analytical procedure. Laboratories should
maintain surrogate control charts using LCSs or MBs results to monitor method performance and
to evaluate surrogate recoveries in actual environmental matrices.
A-62. Target Analyte.
A target analyte is an environmental compound or element that is being measured or identified

in a chemical test to satisfy project-specific data objectives. Target analytes are distinguished
from compounds or elements analyzed solely for the purposes of quality control (e.g., surrogates
and internal standards). For brevity, target analytes are often referred to as analytes.
A-63. Traceability.
Traceability is formally defined as follows: The property of the result of a measurement or the
value of a standard whereby it can be related to stated references, usually national or
international standards, through an unbroken chain of comparisons all having stated
uncertainties (International Vocabulary of basic and general standard terms in Metrology,
ISO, Geneva, Switzerland 1993, ISBN 92-67-10175-1).
A-64. Trip Blank.
Trip blanks are prepared from reagent water and accompany each shipment of aqueous samples
to be analyzed for volatile organic compounds. Analysis of the trip blanks indicates whether
sample cross-contamination occurred during shipment and/or storage.
A-65. Validation.
A number of definitions for the term validation are currently being used in the environmental
testing industry. In this document, data validation or validation refers to a systematic review of
comprehensive data packages, performed external to the data generator, with respect to a
predefined set of technical performance criteria for precision, accuracy, representativeness,
completeness, comparability, and sensitivity. Validation is an objective sample and analyte-
specific evaluation process that involves the application of scientific rather than contractual
criteria to determine whether requirements for a specific intended use are potentially fulfilled.
Validation results in a higher level of confidence when determining whether an analyte is
actually present in an environmental sample at a particular level of interest, but usually results in
a qualitative evaluation of the data. Data validation occurs prior to determining whether the
overall project-specific objectives have been satisfied (i.e., prior to drawing conclusions from the
body of the data).
A-17
EM 200-1-10
30 Jun 05
APPENDIX B
Holding Times and Preservation
Table B-1
Qualification for Holding Times
Flagging for
Preservative 4 Holding Time Holding Time
Parameter Noncompliance
Liquid Solid Liquid Solid Liquid Solid
VOLATILE ORGANICS 12 9 10
Cool 4C Cool 4C, 14 d 14 d Note 1 Note 1
No head space NaHSO4
HCl to pH <28 (aq) to pH
<2
Cool 4oC Note 2
Methanol
Purgeable Aromatic Cool 4oC Cool 4C, 14 d 9 14 d 10 Note 1 Note 1
Hydrocarbons No head space NaHSO4
HCl to pH <28, 12 (aq) to pH
<2
Cool 4oC Note 2
Methanol
Purgeable Halocarbons Cool 4oC Cool 4C, 14 d 14 d 10 Note 2 Note 1
No head space8,12 NaHSO4
(aq) to pH
<2
Cool 4oC Note 2
Methanol
SEMIVOLATILE Cool 4oC 12 Cool 4C 7 d/40 d 3 14 d/40 d 5 Note 2 Note 2
ORGANICS
Benzidines Cool 4C 12 Cool 4C 7 d/40 d 3 14 d/40 d 9 Note 1 Note 1

3 5
Chlorinated Herbicides Cool 4C Cool 4C 7 d/40 d 14 d/40 d Note 2 Note 2
o 12 3 5
Chlorinated Cool 4 C Cool 4C 7 d/40 d 14 d/40 d Note 2 Note 2
Hydrocarbons
Chlorinated Pesticides Cool 4oC Cool 4C 7 d/40 d 3 14 d/40 d 5 Note 2 Note 2
o 12 6
Polychlorinated Dibenzo Cool 4 C Cool 4C 30 d 30 d/45 d Note 2 Note 2
Dioxins & Furans /45 d 6
Explosives Cool 4oC Cool 4oC 7 d/40 d 3 14 d/40 d 5 Note 2 Note 2
o 12 3 5
Haloethers Cool 4 C Cool 4C 7 d/40 d 14 d/40 d Note 2 Note 2
o 12 3 5
Nitroaromatics & cyclic Cool 4 C Cool 4C 7 d/40 d 14 d/40 d Note 2 Note 2
ketones
Nitrosamines Cool 4oC 12 Cool 4C 7 d/40 d 3 14 d/40 d 5 Note 2 Note 2
o 3 5
PCBs Cool 4 C Cool 4C 7 d/40 d 14 d/40 d Note 2 Note 2
B-1
EM 200-1-10
30 Jun 05
Flagging for
o
Phenolics Cool 4 C, Cool 4C 28 d 28 d Note 2 Note 2
H2SO4 to pH < 2
Phenols Cool 4oC 12 Cool 4C 7 d/40 d 3 14 d/40 d 5 Note 2 Note 2
o 3 5
Phthalate Esters Cool 4 C Cool 4C 7 d/40 d 14 d/40 d Note 2 Note 2
o 12 3 5
Polynuclear Aromatic Cool 4 C Cool 4C 7 d/40 d 14 d /40 d Note 2 Note 2
Hydrocarbons
METALS Cool 4oC, Cool 4C 6 months 6 months Note 2 Note 2
HNO3 to pH < 2
Metals except Cr(IV) and
Hg
Chromium (VI) Cool 4oC Cool 4C 24 hours 24 hours Note 1 Note 1
o
Mercury Cool 4 C, Cool 4C 28 d 28 d Note 1 Note 1
HNO3 to pH < 2
MISCELLANEOUS Cool 4oC, NaOH Cool 4C 14 d 14 d Note 1 Note 1
to pH >12 11
Cyanide (Total &
Amenable)
TRPH Cool 4oC, Cool 4C 28 d 28 d Note 2 Note 2
H2SO4 to pH < 2
Acidity Cool 4oC NA 48 hours NA Note 1 NA
Alkalinity Cool 4C NA 48 hours NA Note 1 NA
o
Ammonia Cool 4 C, NA 28 d NA Note 1 NA
H2SO4 to pH < 2
Biochemical Oxygen Cool 4oC, NA 48 hours NA Note 1 NA
Demand (BOD) H2SO4 to pH < 2
Biochemical Oxygen Cool 4oC NA 48 hours NA Note 1 NA
Demand (Carbonaceous)
Bromide None NA 28 d NA Note 2 NA
o
Chemical Oxygen Cool 4 C, NA 28 d NA Note 1 NA
Demand (COD)1 H2SO4 to pH< 2
Chloride None None 28 d 28 d Note 2 Note 2
Chlorine, Total Residual None NA A.S.A.P. NA NA NA
o 12
Coliform, Fecal & Total Cool 4 C Cool 4C 6 hours 6 hours Note 1 Note 1
o
Color Cool 4 C NA 48 hours NA Note 1 NA
o
Common Ions Cool 4C Cool 4 C 28 d 28 d Note 2 Note 2
Dissolved Oxygen, Probe None Required NA A.S.A.P NA Note 1 NA
Dissolved Oxygen, Fix on Site /Store NA 8 hours NA Note 1 NA
Winkler Method in dark
Fecal Streptococci Cool 4oC 12 NA 6 hours NA Note 1 NA
Fluoride None Required NA 28 d NA Note 2 NA
B-2
EM 200-1-10
30 Jun 05
Flagging for
Hardness HNO3 or H2SO4 NA 6 months NA Note 2 NA
to pH < 2
Hydrogen Ion (pH) None Cool 4C A.S.A.P. A.S.A.P. NA
o
Kjeldahl & Organic Cool 4 C, NA 28 d NA Note 2 NA
Nitrogen H2SO4 to pH < 2
Nitrate Cool 4oC Cool 4C 48 hours 48 hours Note 1 Note 1
o
Nitrate-Nitrite Cool 4 C, Cool 4C 28 d 28 d Note 1 Note 1
H2SO4 to pH < 2
Nitrite Cool 4oC NA 48 hours NA Note 1 NA
o
Oil & Grease Cool 4 C, Cool 4C 28 d 28 d Note 2 Note 2
H2SO4 to pH < 2
Organic Carbon Cool 4oC, Cool 4C 28 d 28 d Note 2 Note 2
HCl or H2SO4 to
pH < 2
Orthophosphate Filter NA 48 hours NA Note 1 NA
Immediately
Cool 4oC
Phosphorus (Elemental) Cool 4oC NA 48 hours NA Note 1 NA
o
Phosphorus (Total) Cool 4 C, NA 28 d NA Note 1 NA
H2SO4 to pH < 2
Gross Alpha HNO3 to pH < 2 Cool 4C 6 months 6 months Note 2 Note 2
Gross Beta HNO3 to pH < 2 Cool 4C 6 months 6 months Note 2 Note 2
Radium (Total) HNO3 to pH < 2 Cool 4C 6 months 6 months Note 2 Note 2
o
Residue, Filterable Cool 4 C NA 7d NA Note 1 NA
o
Residue, Cool 4 C NA 7d NA Note 1 NA
Non-Filterable
Residue, Settleable Cool 4oC NA 48 hours NA Note 1 NA
o
Residue, Total Cool 4 C NA 7d NA Note 1 NA
o
Residue, Volatile Cool 4 C NA 7d NA Note 1 NA
o
Silica Cool 4 C NA 28 d NA Note 2 NA
o
Specific Conductance Cool 4 C NA 28 d NA Note 2 NA
o
Sulfate Cool 4 C Cool 4C 28 d 28 d Note 2 Note 2
o
Sulfide Cool 4 C, Cool 4C 7d 7d Note 1 Note 1
4 mL ZnAc plus
NaOH to pH > 9
Sulfite Cool 4oC NA A.S.A.P. NA NA NA
o
Surfactant Cool 4 C NA 48 hours NA Note 1 NA
TCLP Volatile Organics Cool 4C Cool 4C 14 d/NA/ 14 d/NA/ Note 1 Note 1
14 d 7 14 d 7
B-3
EM 200-1-10
30 Jun 05
Flagging for
TCLP Extractable Cool 4C Cool 4C, 14 d/7d/ 14 d/7d/ 40 Note 1 Note 1
Organics No headspace No 40 d 7 d7
Headspace
TCLP Inorganics, Hg Cool 4C Cool 4C 28 d/NA/ 28 d/NA Note 1 Note 1
28 d 7 /28 d 7
TCLP Inorganics, Cool 4C Cool 4C 180 d/NA/ 180 d/NA/
all other metals 180 d 7 180 d 7
Temperature None Required NA A.S.A.P. NA NA NA
o
Total Organic Halogens Cool 4 C, 28 d 28 d 28 d Note 2 Note 2
(TOX) H2SO4 to pH < 2
Turbidity Cool 4C NA 48 hours NA Note 1 NA
Notes:
1. Qualify all nondetections with the R flag and all detections with the J- flag or X flag when the holding time
limit is exceeded.
2. When the holding time limit is exceeded but a gross holding violation does not occur, qualify all nondetections
with the UN flag and all detections with the J- flag. When the holding time is grossly exceeded, qualify all
nondetections with the R flag and all detections with the J- flag or X flag.
3. 7 days until extraction; analyzed within 40 days of extraction.
4. High concentration samples only require cooling to 4C.
5. 14 days until extraction; analyzed within 40 days of extraction.
6. 30 days from collection to extraction; 45 days from collection to analysis.
7. Holding times shown are as follows: From collection to TCLP extraction; from TCLP extraction to preparative
procedure; from preparative procedure to analysis.
8. All aqueous VOC samples must be protected from light. Project-specific preservatives (e.g., mercuric
chloride, copper sulfate, and sodium azide) may be used when matrix interference (e.g., high carbonates)
precludes the use of acid preservation.
9. For aqueous aromatic VOCs, the holding time is 7 days if the samples have not been preserved with HCl or a
project-specific preservative (but have otherwise been properly preserved).
10. If the samples are stored in an air-tight vessel (e.g., the EnCore sampler) at 4oC without taking additional
preservation measures (e.g., methanol immersion and acidification), then the holding time is only 48 hours.
Sample results should be qualified for holding time exceedences as described in Note 2.
11. If oxidizing agents are present (e.g.,residual chlorine), add 0.06 g of ascorbic acid or 5 mL of 0.1 N NaAsO2
per liter of sample.
12. Free chlorine must be removed by the addition of 0.008% Na2S2O3.
d = day
NA = Not Applicable
A.S.A.P = As soon as possible, immediately
B-4

US Army Corps of Engineers - Guidance For Evaluating Performance Based Chemical Data PDF

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

US Army Corps of Engineers - Guidance For Evaluating Performance Based Chemical Data PDF

Uploaded by

Copyright:

Available Formats

EM 200-1-10

GUIDANCE FOR EVALUATING PERFORMANCE-

This manual is approved for public release; distribution is unlimited.

Electronic copies of this and other U.S. Army Corps of Engineers

vPrinted on Recycled Paper

2. Applicability. This EM applies to HQUSACE elements, major subordinate commands,

3. Distribution Statement. Approved for public release and unlimited distribution.

FOR THE COMMANDER:

Subject Paragraph Page

Chapter 2. Data Review Reports

Chapter 3. Data Qualifiers

Subject Paragraph Page

Chromatographic Methods With 2-D Detectors............................................................. 4-2.11.4 4-8

Chapter 5. Holding Times and Preservation

Chapter 6. Data Review Reports

Chapter 7. Initial Calibration

Chapter 8. Initial Calibration Verification (ICV)

Subject Paragraph Page

Qualification................................................................................................................... 8-4 8-2

Chapter 9. Continuing Calibration Verification (CCV)

Chapter 10. Blanks

Chapter 11. Laboratory Control Samples (LCSs)

Subject Paragraph Page

Chapter 13. Surrogates

Chapter 14. Data Review Reports

Chapter 15. Data Review Reports

Chapter 16. Internal Standards for Organic Chromatographic Methods

Subject Paragraph Page

1-1. Opening Remarks.

1-2. Scope and Limitations of Performance-Based Data Review.

a. In general, data packages must contain enough information to evaluate PARCCS.3

c. In order to successfully implement a PB review, the DQOs in project planning docu-

d. In order to perform a PB data review, project-specific data quality objectives must be

e. Performance-based data review must not be performed as a last-minute activity

1-2.1. Performance-Based Data Review Versus Data Validation.

a. This guidance is generally applicable to any instrumental performance-based method,

d. A more thorough data evaluation should be considered when significant QC problems

1-2.2. Performance-Based Data Review Versus Usability Assessment.

a. It is emphasized that the PB data review activities discussed in this document

c. Little or no usability assessment is typically performed during data review or valida-

1-3. Overview of Performance-Based Data Review.

(2) Holding Time and Preservation.

(3) Initial Calibration.

(4) Initial Calibration Verification.

(5) Continuing Calibration Verification.

(6) Sensitivity (e.g., detection and quantitation limits).

(7) Blanks (e.g., field and method blanks).

(8) Laboratory Control Samples.

(9) Post-Digestion Spikes (for trace metal methods).

(10) Matrix Spikes.

(11) Matrix Spike Duplicates and Matrix Duplicates.

(12) Surrogates (for organic chromatographic methods).

Data Review Reports

2-2. Cover Page.

The cover page specifies the following information:

a. Unique report ID number.

b. Name and address of data reviewer.

d. Client name and address.

e. Project name and site location.

f. Statement of data authenticity and official signature of release.

2-3. Cover Letter.

a. Project name (or brief description of the project).